Language policy formation in higher education: discursive tension and legitimacy-seeking deliberation

ABSTRACT How are processes of language policy formation carried out? Besides, how does one negotiate competing discourses in a policy process to ensure its legitimacy? Such questions sparked the interest for this study, which examines legitimization processes involved in the development of language policy guidelines at a higher education institution in Norway. The article examines how different actors at the institution strive to obtain legitimacy for their different views, as well as for the policy process as a whole. The data collected include audio recordings of meetings in a language policy committee, submissions from a round of consultations, and language policy guidelines. The article shows how actors drawing on different discourses concerning use of English and Norwegian met during the process. Due to tension between discourses, the policy committee had to take several steps in order for the process and the final guidelines to be acknowledged as legitimate. Most importantly, a range of voices and discourses had to be included. The article offers insight for language policy scholarship as it highlights policy formation in higher education settings and draws attention to how the context of policy formation defines what is seen as legitimate, allowing for future comparative perspectives.


Introduction
Legitimacy, a discursive sense of acceptance (Vaara et al., 2006), is essential if a language policy is to be respected; the question is how it can be obtained. This article follows the process of formulating language policy guidelines for the Western Norway University of Applied Sciences (HVL 1 ). In the article, the discussion concerning the extent to which English should be used at the institution serves to illustrate the need for legitimacy in the process and what actors do to obtain it. The specific research question asked is: How do the different actors at HVL strive to obtain legitimacy for their views and for the policy process as a whole? According to Hult, most studies in language policy and planning (LPP) have looked at policy texts or the implementation of policy, while " [r] elatively few LPP investigations have examined the making of policy texts in real time" (Hult, 2015, p. 226). Many studies look at policy for primary and secondary schooling, rather than for higher education, and it is more common to study implementation (e.g., Johnson, 2010) and results of policy (e.g., Gándara & Hopkins, 2010) than actual policy formation. For studies where policy creation in higher education is the key focus, see e.g., Källkvist and Hult (2016) and Miranda et al. (2016). By studying an entire policy making process, this article contributes with new understanding of the development of language policy. More importantly, it aims to contribute with new insight into the role of legitimacy in such processes. In the history of language policy, principles of authority have been more important than principles of democratic participation (e.g., Wrigth, 2004). However, Ramberg (2016) argues that today, democratic ideals are essential in such processes, and a wide range of voices and discourses must be engaged in the process. This study illustrates how legitimacy is interlinked with the policy process, and the crucial role legitimacy may play.
In order to answer the research question presented above, I look at the discussions in the committee that drafted language guidelines for HVL and at the comments on the guidelines given in a round of consultations conducted at the institution. The round of consultations became a site of tension between actors drawing on different discourses, who expressed conflicting views and interests. The findings show that several steps were taken in order for the policy process as a whole, and its result, to be seen as legitimate.
I start out by providing a brief account of language policy and practice in Norwegian academia, as this serves as the backdrop for the study. Then, the institution HVL will be presented, followed by the study's theoretical and methodological framework. I then proceed to the findings, and lastly, I discuss legitimacy in the language policy process and connect the pursuit for legitimacy to the hope for exemplary practice.

English in policy and practice in Norwegian higher education
The use of English in academia is a topic that has created heated debates over the last years in Norway, as in many other countries. Numbers show that the use of English is increasing in all parts of higher education. In 2016, English was the language of instruction in 19.6% of all academic study programmes, compared to 8.9% in 2007 (Schwach & Elken, 2018, p. 19). English is also the most used language in PhD theses and scientific publications.
Based on such statistics, one may argue that Norwegian is under pressure in academia. Therefore, governmental language policy states that Norwegian should be the main language in higher education, and English only used when needed (Kulturdepartementet, 2008). However, national policy documents on research convey competing signals. A white paper on research (Kunnskapsdepartementet, 2013) highlights the need for increased internationalization, and thus indirectly argues for an increased use of English. The two policies are not in line with one another, and institutions are left in a difficult position. The ideological tension found in policy regarding higher education is not unique to Norway. Hultgren et al. (2014) distinguish between two opposing discourses associated with Englishisation of Nordic universities: "internationalist" discourse and "culturalist" discourse. Safeguarding of the national language is a key part of the "culturalist" discourse. Contradictory discourses that center around nationalizing and globalizing orientations are also found in other studies (e.g., Salö, 2016;Soler & Vihman, 2018). The data in this study can provide insight into how policy actors deal with competing discourses and ideological tension.

Western Norway university of applied sciences
HVL was established on January 1 st 2017, as a result of a merging of three different university colleges. 2 It is a national, public institution that has five campuses spread out along the west coast of Norway. It is one of the largest higher education institutions in the country with approximately 17,000 students and 1 600 staff members.
The merger agreement between the three institutions states ambitions, profile, and goals; one example being to achieve university accreditation. Internationalization is mentioned several times in the agreement, most explicitly at page three: [HVL] shall be proactive internationally through binding schemes for mobility in all educational cycles [bachelor, master, PhD], through flexible arrangements for employee mobility, and be visible internationally through increased external funding of education and research (merger agreement, p.3, my translation) It is clear that HVL has the ambition to become an international institution. However, it is not possible to determine to what degree, nor what linguistic implications this goal might have. At the same time, the institution aims to have strong local ties and be an important actor in Western Norway, as illustrated by this sentence from the agreement: "HVL shall be the preferred partner for cooperation and shall leave traces in the development of the West-Norway region" (merger agreement, p. 2, my translation). Thus, the two opposing discourses presented above, "internationalist" and "culturalist," are present in HVL's policy documents. In this article, they will be referred to as "the value of internationalization" and "safeguarding of Norwegian." In the spring of 2017, a language policy committee was established to draft HVL's language policy guidelines. The committee had the following mandate: The committee shall develop a proposal for language policy guidelines for HVL. The work shall build upon [a number of internal policy documents & university legislation]. The committee will consist of one representative from each region, plus one student. (Mandate, my translation) The committee and the consultative round the guidelines went through will be presented in the section Data and methodological framework.

Language policy processes and social actors
In order to capture the broad scope of language policy (LP), this article adapts Johnson's (2013, p. 9

) definition:
A language policy is a policy mechanism that impacts the structure, function, use, or acquisition of language and includes a) Official regulations [. . .]  The focus of the study presented in this article is on official policy, and both the process of policy formation and the final product is studied. However, the data I have obtained by closely following the development of language guidelines, as well as my interest in legitimacy in policy processes, speak in favor of emphasizing the process of policy formation.
Social actors are important in LP processes, both as groups involved in negotiation, and as individuals with power to influence these processes. Johnson and Johnson (2015, p. 226) claim that "policy decisions are socially negotiated between multiple actors within and across levels," and that some actors -language policy arbiters -can have "singular power with regard to how a policy is interpreted and appropriated and all subsequent decisions in the policy process must funnel through them." In the LP process at HVL, former and current policies are interpreted and appropriated by the LP committee. Thus, the committee members operate as policy arbiters, especially in moments when they intervene directly and make distinct choices regarding the policy.

The pursuit of legitimacy
It is well-known that policy is often far from practice, and this article sees legitimacy as central for policy to be respected by its intended users. I define legitimacy as "a discursively created sense of acceptance in specific discourses or orders of discourse" (Vaara et al., 2006, p. 789). Legitimacy cannot be seen independent of the particular context, as the emergence of discourses always takes place in a socio-historical context (Vaara et al., 2006, p. 793).
My position on legitimacy is influenced by Ramberg, a Norwegian philosopher of language, who in his paper on deliberative legitimacy in language policy writes the following: Legitimate decisions are central to the democratic mediation of this basic opposition between divisive conflict of view or interest and the shared need for common, coordinated action. Legitimacy ensures that concerted action is shaped and executed not principally through force, but by virtue of acknowledgement. Thus, the central function of political legitimacy is to confer a normative stamp; legitimate decisions-in-conflict ought to be respected. (Ramberg, 2016, p. 65, italics in the original) Thus, legitimacy is seen as essential for the result of LP processes to be accepted. I also build on the notion of deliberative democracy. According to Dryzek, deliberative democracy requires that for a collective decision to be legitimate; it must be subject to reflective acceptance of those subject to it, who should be able to participate in consequential deliberation concerning the production of the decision. (Dryzek, 2010, p. 37) Moreover, it has been argued by some scholars that democracy needs to be understood as consideration of arguments and deliberation between competing discourses, not by counting numbers. In this study, the Norwegian society constitute the context and discourses circulating here will define what is seen as legitimate. In Norway, questions of democracy and legitimization are at the core of any policy process. It is not unusual for actors that are dissatisfied with a final policy to attack the policy process for not being democratic, in order to undermine the legitimacy of that policy (see e.g., Røyneland, 2016). Thus, the democracy discourse is dominant and needs to be drawn upon to discursively create legitimacy and acceptance.
Legitimation provides explanations and justifications of specific practices (see Berger & Luckmann, 1966;Van Leeuwen, 2007). Legitimation strategies, ways in which legitimacy can be created, can be conscious or unconscious (see e.g., Van Leeuwen, 2007). Van Leeuwen's (2007, p. 91) claims that discourses construct legitimation for social practices, and he distinguishes four key categories of legitimation: authorization, moral evaluation, rationalization, and mythopoesis. As will be evident, authorization and rationalization are strategies or tools drawn upon in the policy process at HVL, as they provide explanations and justifications for the actions taken.
In this article, my concern is not with the legitimacy of one language or variety (see e.g., Camps, 2017), or with the legitimacy of speakers. Rather, it is with how legitimacy is created in LP processes in order for a policy to be acknowledged as legitimate. According to Johnson (2009Johnson ( , 2013, the multiple layers of LP is characterized in terms of processes; creation, interpretation, and appropriation. I claim that legitimization is crucial throughout all the three stages. Further, I claim that gaining the recipients' accept of the creation of a policy may contribute to legitimacy in the two following stages as well. Thus, legitimacy is interlinked with the policy process itself. The analysis thus deals with how legitimacy is sought on three steps: the deliberations in the committee, the process as a whole, and for the final guidelines.

Data and methodological framework
The data in this study can roughly be divided into two parts: 1) data from the work of the language policy committee before the round of consultations, consisting of audio recordings and field notes from three meetings, all e-mail exchanges in the committee, a draft of an accompanying letter for the language policy guidelines and several drafts of the guidelines, and 2) data from the round of consultations, consisting of consultative submissions, the committee's discussion of these, as well as the final guidelines. The whole data set is presented in Table 1.
As evident from the table, it is an extensive material. For this article, the main material is part two of the data (marked in italics in the table), from the work done in the language policy committee following the round of consultations, as well as the consultative submissions themselves. I will draw on the rest of the material where relevant. The study also takes account of national and international policy documents. Table 2 provides an overview of documents drawn upon by the actors in the LP process at HVL.

The language policy committee
The committee established to draft HVL's language policy guidelines consisted of five members: a senior advisor, two scientific staff members from the Department of Language, Literature, Mathematics and Interpreting, an administrative staff member (Office for Communications and Society contact) and a student. These members were chosen in order to ensure that all the three former institutions (HSH, HiSF and HiB) were represented. Further, the composition of people ensured that different groups (administrative, scientific, student), and both genders, were represented. The committee had three meetings; two in person and one through Skype. The student did not take part in the first meeting, as they had not yet been appointed by the Student Parliament.

The round of consultations
When the committee had come to a consensus regarding the draft of the language policy guidelines, the document was distributed at the institution for a round of consultations. The consultative bodies were deans, division managers and the president of the Student Parliament. The remarks were submitted through an online survey that asked for comments on the guidelines' five different sections (see Discorse encounters: the value of internationalization and safeguarding of Norwegian). Additionally, it was possible to add "Other or overall comments on the proposal for language policy guidelines" (my translation) at the end of the survey. The final number of submissions was low (3 in total), but due to the scope of the submissions, they provide data for a thorough analysis. The three consultative bodies that responded to the round of consultations were the deans of the Faculty of Health and Social Sciences (FHS) and the Faculty of Education, Arts and Sports (FEAS), on behalf of their faculties, and the president of the Student Parliament. In addition to filling in the online survey, FEAS submitted two full-length guidelines with comments in the margins, which the dean had received from employees at the faculty. The committee went through the comments from the survey and deliberated over them and then made some changes to the guidelines. After this, the documents from the committee were forwarded to the University board. The University board approved the guidelines. See Figure 1 for an illustration of the process, and Table 3 for an overview of actors involved in the LP process at HVL. Throughout the whole process, I received all the e-mails with the attached documents that circulated between the committee members. This included all the consultative submissions.

Analysis and researcher positionality
I draw on the ethnography of language policy framework in the analysis, taking into account agents, goals, processes, discourses, and the social and historical context (Johnson, 2009). According to Johnson (2009, p. 140) "ethnography of language policy [. . .] compares critical discourse analyses of language policy texts with ethnographic data." The combination facilitates a multidimensional investigation (see Källkvist & Hult, 2016) and illuminates the complexity in LP processes. In line with Källkvist and Hult (2016) analysis of discourse processes in policy formation in Sweden, I apply discourse analytic strategies to examine the data in this study. I observed and audio-recorded all the meetings in the LP committee. Afterward, I listened to the audio recordings and took detailed minutes describing what was taking place. This made it possible for me to keep track of the content of the meetings, and later to go back to various events in the recordings. The minutes, as well as all the consultative submissions, were then coded. After the initial coding, the data was recorded, and the codes grouped together in categories based on similarities. All the codes were generated from the data. Later, the parts of the data that caught my interest as relevant for the research questions were transcribed. The data from the committee meetings presented in this article resulted from these transcripts. I have translated all excerpts from Norwegian to English.
In the initial descriptive coding, the topic "English" occurred repeatedly, both in the detailed minutes from the committee meetings and in the consultative submissions. Since competing interests regarding use of English and Norwegian were apparent from the data, I wanted to look more closely at the discussions regarding use of English and how the committee dealt with competing interests. Hence, most excerpts from the data presented in the article are chosen based on that they were coded with the topic "English" or "Norwegian/National language." Other excerpts from the data were coded with "Intertextual/interdiscursive connection." By employing interdiscursivity (Fairclough, 1992), a concept closely related to intertextuality (Kristeva, 1986), particular trajectories of topics and discourses are traced backwards to the work in the LP committee and forward to the final version of the guidelines. Interdiscursivity "draws attention to the potential heterogeneity of text in terms of the diverse discourse conventions, types of discourses, which can be drawn upon in their production" (Fairclough, 1992, p. 284). Language policies may be linked to past and current policy documents, as well as to past and present discourses (Johnson, 2015). This article makes visible these discourse connections. In the article, I look for measures taken throughout the process to ensure that the language policy process and the final guidelines are acknowledged as legitimate. As will be clear, interdiscursivity is a central part of the legitimization process.
When access is granted to study language policy processes, the researcher is often part of and a key participant in the committee creating the policy (e.g., Källkvist & Hult, 2016;Røyneland, 2016). This was not the case for this study. The access was gained by strategically establishing contact with key actors and by a stroke of luck. Prior to the committee being established, I had been in contact with several people, including the person who later was asked to lead the committee, regarding policy issues at the institution. This person therefore knew that I was interested in the policy process and was willing to let me observe the meetings in the committee. Without my knowledge, they also informed the other committee members about this. Then, I contacted the members myself and was granted access. During the meetings, I only observed; however, I took part in informal conversations during lunches and coffee breaks. I intended to position myself as a researcher interested in language policy, without sharing personal opinions on language related issues. However, the fact that I come from the west coast of Norway, am a user of Nynorsk -the main langauge at HVL -and work in academia, means that I in many ways can be seen as an "insider." It is likely that this was in my favor when I was granted access to observe the LP process at the institution. Furthermore, the committee members were likely to know that I had been involved in language activism for Nynorsk (see Thingnes, Thingnes, 2020a, Thingnes, 2020b for an analysis of the discussions regarding Nynorsk in the policy process at HVL and researcher positionality).

Findings
When presenting the findings, the main focus is on the consultative submissions and the policy committee's responses to the submissions. I start off by presenting the discourses that met in the consultative round, before turning to the committee's responses to the consultative submissions. Then, I turn to the role of interdiscursivity in legitimizing the different stances taken by the consultative bodies and the committee.

Discourse encounters: the value of internationalization and safeguarding of Norwegian
The language policy guidelines commented on in the round of consultations (Proposal 27.11.2017, from now on referred to as "guidelines") have five sections: 1 Overarching language policy, 2 Education and teaching, 3 Research and dissemination, 4 Administrative matters, and 5 Follow-up and language quality measures (my translations). The second section got most attention from the consultative bodies. When the Chair had read all the consultative submissions, they expressed the following in an e-mail to the committee: "As I had expected, the round of consultations came to be about English. This is especially true for education, section 2 in the guidelines." The Chair further summed up saying that most of the statements asked for guidelines that were "more liberal towards use of English" (e-mail, 09.03.2018). One such statement is the following given by the Student Parliament: In the first paragraph, or as an additional paragraph below, it would be desirable to include a sentence about an increased number of courses in English. HVL has stated a goal of receiving more exchange students, and one must facilitate this in the form of expanding our English course portfolio. One can also look at this as a general comment on the section; there is a strong focus on the fact that teaching will be conducted in Norwegian, especially at bachelor and master's level. Here, the desire to include more international students in bachelor and master programmes is not taken into account. (Online survey, Student Parliament) The statement sums up the submissions regarding English quite well, including the arguments presented for an increased use of the language. Three different points of views were aired, the first is related to the need to cater for exchange students, the second related to the perceived need to strengthen the Norwegian students' English proficiency, and the third reflected the view that the guidelines were not in line with other important policy documents regarding the use of English. However, the committee had a different view. In an attachment to the e-mail referred to above, the Chair stated that the committee wished to "strengthen the national language in higher education" (e-mail, 09.03.2018). This indicates that competing discourses met during the round of consultations. A range of discourses were articulated, but two were particularly salient: the value of internationalization (and thus English) versus safeguarding of Norwegian. The presence of two competing discourses is not surprising, taking previous research into account (e.g., Hultgren et al., 2014;Salö, 2016;Soler & Vihman, 2018). In this article, the focus is not on the conflicting discourses. Rather, they are the starting point and the reason to look for measures taken in the pursuit for legitimacy.
It is important to note that while relating to the two discourses, different policy actors took up different positions. Additionally, the consultative bodies did not explicitly express negative views about the use of Norwegian. Two of the consultative bodies argued for the importance of the use of Norwegian as a scientific language in publications in their comments to the guidelines. Nevertheless, the three consultative submissions all, to some extent, called for more use of English in teaching and examinations. Thus, a wish to open up for international students and give local students English proficiency -to be international -turned out to be as important as the use of Norwegian.

Incorporate or dismiss suggested changes?
As illustrated above, the committee and most of the consultative bodies disagreed on the use of English at HVL. The question that arose among the committee members was how the committee should respond to comments that were not in line with its views. As an observer, I was equally interested in the above question. Furthermore, I was interested in seeing if the committee's response could be understood in light of a pursuit for legitimacy. In an e-mail to the committee when the round of consultations had ended, the Chair wrote: We wish to strengthen the national language [Norwegian] in higher education and research, in line with what is stated in the unanimous Nordic recommendation "More parallel, please". The Language Council in Norway endorses this. We must uphold it. Yet I suggest complying with some of the comments here, as well. (Email, 09.03.2018) In their response to the e-mail cited above, the committee members supported the Chair in pursuing the initial wish to safeguard Norwegian, by explicitly expressing their support, and by commenting on specific excerpts from the consultative submissions.
In line with the Chair's e-mail, the committee only incorporated some of the comments from the consultative bodies. For example, the consultative bodies suggested to change from "Norwegian" to "English" in the following sentence in the guidelines: "As a rule, the examination language at master's level should be Norwegian." The committee did not make this change. Thus, its members acted as policy arbiters, going against wishes expressed in the round of consultations, with the outcome that Norwegian was given higher priority. However, the committee did also partly open up for English, as evident from this bullet point that was added to section 1 of the guidelines as a result of the round of consultations: HVL wishes to cater for more exchange students from non-Nordic countries. This means that HVL in both bachelor and masters programmes must have a course portfolio that includes courses in which the teaching takes place in English. (Language policy guidelines HVL) This is where the comments from the consultative bodies regarding English has had the strongest influence on the final guidelines. In some cases, as here, deliberation led the committee members to take up new and slightly different positions. Committee member 3 expressed directly that when it comes to English, "we benefit from being shaken up a bit." I interpret this to mean that the committee thought it benefited from being challenged when it came to the use of English. It also shows that this particular group of actors took a variety of positions toward the use of English, not solely for or against. Toward the end of the article, I see the inclusion of suggested changes as discursive interaction which may contribute to legitimacy in the policy process.

Pursuing legitimacy through interdiscursivity
This section focuses on how the actors in the policy process draw on policy documents when arguing for and legitimizing their views. First, I will present data from the consultative submissions. Then, I will move on to the role interdiscursivity played when the policy committee deliberated over the suggestions from the consultative bodies.

The consultative bodies
The two excerpts below illustrate how the consultative bodies drew on policy documents when arguing for English: The strong focus on and the importance of internationalization is prominent in key documents such as the Status Report for Higher Education in 2017, HVL's annual report for 2016-2017, and the merger agreement. [. . .] The language policy guidelines for the use of English in bachelor and master programmes do not reflect the objectives of the mentioned policy documents, nor do they reflect HVL's focus on English as a parallel language. (Online survey, FEAS) It is also proposed that as a result of HVL's international ambitions and the Ministry of Education's demands for internationalisation, both at home and abroad, one should, in some cases, reverse the mindset that "As a rule, the examination language at master's level should be Norwegian." (Online survey, FEAS) The first statement refers to "key documents" and names three. One of them is external to the institution (a governmental document), while the other two are internal. The documents are explicitly referred to. The reference in the second statement is different. Here, documents are not mentioned as such, but both external (Ministry of Education) and internal ambitions and demands (HVL) are referred to. The fact that the consultative submissions drew upon past and current policy documents, demonstrates the importance of these documents.
The references included in the consultative submissions above also say something about which texts had authority for the consultative bodies. As Lemke (1995, p. 19) claims "[w]e make sense of every word, utterance, or act against the background of (some) other words, utterances, acts of a similar kind." This implies that it is important to "understand just which other texts a particular community considers relevant to the interpretation of any given text" (Lemke, 1995, italics in original). For the consultative bodies the key documents when reading section 2 of the draft guidelines seems to have been those related to external and internal policy on internationalization.
The consultative submissions cited above build on strong discourses on internationalization and, partly following from it, a discourse on English as a valuable language in which international academia operates. These discourses are found in past and present policies, as in the White Paper on research mentioned earlier (Kunnskapsdepartementet, 2013), and they are drawn upon and recontextualised in the consultative submissions, as seen through the reference to the Ministry of Education in the data excerpt above. The consultative bodies did not just want more English, they could point to why and give reasons for their wishes. By drawing on broader circulating discourses, they ensured legitimacy for their views.

The social actors in the committee
After the round of consultations, the Chair sent out a revised version of the guidelines to the committee members, including explanations and questions regarding the changes made. In his overall comment on section 2 Education and teaching, the Chair wrote: The comments on this section [. . .] challenge us greatly when it comes to the use of English. Maybe we have been too restrictive? Probably not, when we see our proposal in conjunction with our accompanying letter. To strengthen and facilitate the use of the national language (Norwegian) in teaching and research is central to our recommendations. This is consistent with the developments in the rest of the Nordic countries (and Europe). I do not think we should depart from this, more than necessary . . . (Comment on Revised guidelines,9.03.18) Here, the Chair expresses a hint of doubt, asking if the committee has been "too restrictive." This is another example of the range of stances taken by individual committee members with regard to the main discourses salient in the round of consultations. Still, the Chair concludes by holding on to the importance of safeguarding Norwegian, and the other committee members supported this in their responses. When looking at interdiscursivity, we see that the two last sentences in the excerpt are of greater importance. Here, the Chair seeks legitimacy discursively by drawing on discourses found in language policy at the local and Nordic level. Most of the documents the committee draws on are referred to in the accompanying letter. They include More parallel, please!, Mål og meining, two reports on language status, as well as language policy guidelines for the institutions that merged to form HVL (see Table 2). What is evident when studying the documents the committee draws on, is that most of these are policy documents on language. These documents are cited in an attempt to legitimize the committee's reaction to the consultative submissions, as well as the final guidelines. There are also other instances where the committee dismisses suggestions from the round of consultations by referring to key policy actors and experts. In an e-mail sent after the round of consultations, committee member 3 wrote the following: "I think we should use terms like parallel language. It is what is being used." (e-mail, 13.03.18) The last part of this statement argues for the use of the term "parallel language use" in the guidelines -instead of "parallel use of Norwegian and English" as suggested in one of the submissions from the Faculty of Education, Arts and Sports -by pointing to the term's wider use. Based on the discussion in the committee, it is likely that the particular committee member referred to the fact that the term is used in national and Nordic policy documents and policy discussions regarding language. The committee member invoked broader discourses on language identified in policy discussions at many Nordic universities (see e.g., Hultgren, 2014;Holmen, 2017) in order to legitimize the committee's actions. The act of dismissing the suggestion from the round of consultations was seen as legitimate by the committee because the term had been used by others. Further, the selection of documents the committee draws on demonstrates the nature of texts that were considered relevant by the committee (cf. Lemke, 1995), which differed from the texts considered relevant by the consultative bodies. This illustrates how the policy process could result in a document constituted by a combination of discourses.
To understand why these particular documents were relevant for the committee, it is necessary to look at its individual members. Two of them are linguists. Furthermore, all of the members, apart from the student, had been involved when their former institutions produced language guidelines. Thus, they have experience of similar work and are likely to have had insight into policy discussions on language in higher education. The committee had some expert knowledge, and it also drew on other forms of expertise: When the Chair argued for the committee's stance on strengthening the position of Norwegian, they pointed to the Norwegian Language Council's endorsement of the policy presented in More parallel, please!. This institution must be seen as the one having most authority when it comes to language (policy) in Norway. According to Blackledge (2006, p. 126), legitimation of texts can be gained through references to a person or an institution in a highly respected position. Thus, referring to the Language Council could be a step toward ensuring legitimacy. Authorization is also part of Van Leeuwen's (2007) four categories of legitimation, and it is sought through reference to e.g., "persons in whom institutional authority of some kind is vested" (Van Leeuwen's, 2007, p. 92). Van Leeuwen (2007) points to the fact that several forms of legitimation can occur in combination and he highlights that the categories also can be used to de-legitimize. In the case of the Language Council, authority is connected to expertise. However, as Røyneland (2016) illustrates, in Norway expertise is not sufficient for a language policy to be accepted. On the contrary, broad societal representation is an as central criterion and being an expert may, in fact, be seen as de-legitimizing as experts are often conceptualized as people who are out of touch with realities. The fact that several of the committee members were language (policy) scholars, could thus potentially have led to critique of the committee's work. Yet, the composition of the committee ensured a different kind of representation, namely geographical. It is likely that this was as important for the intended users of the guidelines at HVL. This is due to the institution being a result of a recent merger of three different institutions, with diverse interests that had to be taken into account in the policy process.

Discussion
As mentioned in the introduction and touched upon above, democratic ideals are essential in policy processes on language in Norway today. Due to this, and due to the fact that the committee was mainly composed of scholars with special knowledge of language issues -who drew on language policy documents -the policy process itself had to ensure that other voices and discourses were engaged. If the committee wanted the policy to be accepted and followed, despite its rejection of the wishes expressed by the consultative bodies, it was essential for the process to meet the ideals of deliberative democracy.

Legitimacy through discursive representation
Following Ramberg (2016), I see discursive representation as a key to achieving legitimacy in a policy process. There was a tension between two main discourses that met in the construction of language policy at HVL: the value of internationalization, and thus English, and the wish to safeguard Norwegian. While drawing primarily on one of the two discourses, the social actors involved in the process took a range of stances. In order to achieve legitimacy, the committee had to deliberate across competing discourses. The question thus became whether the committee in the process was able to incorporate different discourses to a sufficient degree and "bring different discourses to bear on the same issue" (Ramberg, 2016, p. 75). If the committee had not been able to do so, there would not have been sufficient discursive representation. However, consensus is not always possible. What is important, highlights Ramberg, is that there are "traces of dynamic discursive interaction and change" (Ramberg, 2016, p. 76). Such traces are seen in the process at HVL in two ways. First, when suggestions were dismissed -often due to a conflict between discourses -the committee provided reasons in their deliberations. Second, the committee made changes to the guidelines as a direct result of the round of consultations. The latter demonstrates that the committee was willing to accommodate, adjust, and revise features "of any particular discourse so as to make an increasing number of connections and cross-points and calibrations with competing discourses" (Ramberg, 2016, p. 76). Overall, when the committee exercised political authority, it could account for its decisionmaking and show how several discourses influenced the committee's decision making.
The final stage of the LP process was approval by the University board (see The round of consultations and Figure 1). This can be seen as a central condition for legitimacy since it demonstrated the importance of the guidelines and the fact that they resulted from a political process. Additionally, it ensured that people in positions of authority were familiar with the guidelines, which could facilitate implementation. The committee saw this as an important step toward setting guidelines that would actually be respected.

Why pursue legitimacy? Aiming toward practice
According to the Norwegian Language Council, language policy guidelines at Norwegian universities are barely known among students and employees and are only integrated into an institution's main strategy to a small extent (Larsen & Lie, 2018). If this is the case, why pursue legitimacy? Can ensuring legitimacy for LP processes lead to a situation where language policy guidelines are known and respected? Cooper (1989, p. 75) claims that "planners need to consider how they will persuade potential adopters to accept the innovation." He further argues that planners have to act as marketers and thus must "consider ways and means to create a climate of opinion favourable to the adoption of the planned innovation" (Cooper, 1989, p. 78). At HVL, the LP guidelines are the innovation. Ensuring that the policy process is legitimate may lead to it being accepted. Legitimization is thus one potential means to persuade potential adopters of the policy product and foster satisfactory appropriation. HVL's final guidelines state that the institution aims to be a "linguistic role model for Norwegian universities." When analyzing the data, it became evident that legitimacy was essential in ensuring that the guidelines fulfil the committee's aims regarding content, reception, and implementation. Røyneland (2016) demonstrates how language users can react negatively to policy processes even though all steps have been taken to ensure a democratic process. However, at HVL, at this stage, there have to my knowledge not been any negative reactions following the approval of the final guidelines, which suggests that the policy process is acknowledged and the result is generally seen as legitimate. As expressed by the Chair, one of the reasons for carrying out the round of consultations was to raise awareness of the guidelines across the institution. Additionally, users of the guidelines had an opportunity to "make their claims count" (Ramberg, 2016) and they may perhaps feel more obliged to follow the guidelines as a result. However, there were few consultative submissions, meaning that only a small number of social actors seized the opportunity to express their opinion, and silence does not necessarily mean support -it could as well be seen as a lack of interest for language policy. As Haugen (1966, p. 24) famously said "The planner proposes, but the community disposes." It is too early to conclude to what extent the policy process at HVL has resulted in guidelines that will be respected, and if HVL will become the "linguistic role model" it aims to be. Further research could tell us more about the extent to which employees and students are familiar with the guidelines, how the guidelines are perceived by different actors, and how they are interpreted and appropriate at the institution.
The policy process that took place at HVL is not unique to this institution. Processes of this kind have become regular practice in Norway (see e.g., Røyneland, 2016). Linn (2010, p. 126) claims that "[t]he Norwegian experience [. . .] has been that peaceful language policy can only be achieved in a democracy once the voice from below is heeded." The social actors developing policy must listen to the potential users of that policy. Therefore, round of consultations are common in Norway today, and a variety of actors and a range of voices are involved in processes regarding language policy issues. Seen from the outside however, the process at HVL might be perceived as unusual. Van Leeuwen (2007, p. 95) claims that "expert authority may be waning, albeit only slowly." There are still policy processes in the world that mainly seek legitimacy through expertise or authority and that do not involve the intended users of a policy. What is seen as legitimate, is shaped by the context of policy formation and salient discourses. Therefore, the context needs to be taken into account, but if democratic ideals are important, the formation of language policy guidelines at HVL can serve as an example of how a policy process can be undertaken, and this study can allow for a comparative perspective on policymaking.

Summary and conclusion
In this article, I have asked how different actors at HVL strived to obtain legitimacy for their views and the policy process as a whole. The value of internationalization, and thus English, and safeguarding of Norwegian are two discourses that are articulated in national and local policy documents on research and higher education, and on language, respectively. The different social actors in the policy process at HVL drew on these documents when arguing for their positions, demonstrating the important role of national policies, as well as the tension in policy regarding higher education.
Democratic actions are essential for a policy to be accepted by its users. Because of discursive tensions between the committee proposing the language guidelines and the consultative bodies, the need for discursive representation in the process was pressing. In order for the process to be acknowledged as legitimate -and thus contributing to legitimization of the final guidelines -the committee needed to show that a range of voices and discourses were engaged. The round of consultations in itself was the most important step in this respect, additionally the committee members did deliberate over the consultative submissions and thus kept all salient discourses and relevant positions in the policy debate within reach. The final guidelines can be seen as a compromise and a product of many voices, discourses and policies, or, in Bakhtin's words, they have a "background made up of contradictory opinions" (Bakhtin, 1986, p. 281). However, the committee had the last say.