Applying transnational strategies locally : English as a medium of instruction in Swedish higher education

Analysing student interaction at an international u niversity in Sweden, this article investigates how the transnational strategy of usin g English as a medium of instruction can be (re)constructed by students in a loc l context. The analysis focuses on students’ expectations—or norms—for when it is appropriate to use English, and shows that English medium-instruction does not necessarily mean that students speak English all the time. The local language Swedish is used in connection to teaching and students establish local norms for when, how and with whom it is appropriate—or inappropriate—to speak En glish. A conclusion is that although language choice at the international unive rs ty is influenced by global factors, it is still firmly a local construction.


Introduction
The current processes of globalisation bring about new meeting places for people from different countries and parts of the world.More people than ever live and work in places other than where they were born and the number of students on exchange is increasing.As Appadurai (1996) notes, we live in a "world of flows", where ideologies, peoples and goods, images, messages and discourses move across national boundaries and societies.The world of flows affects social relations and, obviously, "[g]lobalisation is proving to be the salient context for an increasing number of local sociolinguistic experiences" (Coupland 2003: 466).
In this article, I examine how the transnational strategy of using English as a medium of instruction can be (re)constructed in a local context influenced by global flows, namely six international university courses in Sweden.The number of courses in English has increased in Europe as well as in Asia, as part of universities' strategies for internationalising higher education (Ammon andMcConnell 2002, Wächter andMaiworm 2008).The official choice of language is thus influenced by transnational strategies and flows, and the international university seems to be a fruitful setting for studying how actors can handle global flows in local contexts.
I begin with an overview of the international university and how universities function as meeting points for global and local, social and linguistic processes.The overview places the study in the field of sociolinguistics of globalisation and theories and studies presented by Appadurai (1996), Pennycook (2007) and Blommaert (2010).I then give an account of the empirical data and how the transnational method of using English as a medium of instruction can be studied in a local context through analyses of students' language choice in video-recorded classroom interaction.I focus on norms for language choice and how students can construct norms for when it is appropriate-or inappropriate-to use English.The analysis takes its departure in sequences where students or teachers orient towards a certain linguistic system as an instance of deviance and, thereby, simultaneously make visible norms for language choice (Gafaranga 2000).I also discuss participants' attitudes to varieties of English and to the linguistic context.A conclusion that I will come back to is that language choice at the international university is "obviously influenced by global factors, but still firmly local" (Blommaert 2010: 180).

The international university
Universities have always been part of an international arena.Throughout history, researchers have collaborated across institutional and national borders, and students have travelled abroad for education.In recent years, however, universities have made a point of emphasising the global character of higher education and formulated strategies, methods and goals for internationalisation.
The strategies, methods and goals are often shared and the internationalisation of higher education appears as a phenomenon characterised by transnational influences and flows.For instance, a reoccurring goal of internationalisation in higher education is that students should learn about others' experiences and world views, be trained in cross-cultural communication and gain a broader perspective on their own culture (Otten 2003). 1 Mobility and exchange of students and staff is a method used, and meetings across cultural and national boundaries are made possible through the introduction of a global contact language, notably English (Coleman 2006).In this way, courses and study programs in English are part of the process of internationalising higher education, and-just as with other formal aspects of internationalisation-the choice of language is imbued with influences from other national and international contexts.Rather than any other contact language, it is English that is the official medium of instruction across national boundaries in Europe and Asia (Ammon andMcConnell 2002, Wächter andMaiworm 2008).
The global spread of cultural forms and languages has been discussed in terms of homogenisation of world culture, Americanisation and media imperialism (see e.g.Phillipson 1992).However, empirical studies at the international university demonstrate that the linguistic environments taking shape are complex and that discussing them in terms of hegemonic lingua franca use is insufficient (Haberland et al. forthcoming).Knight (2004) points out that the internationalisation of higher education means different things to different actors in the field, and Roberts (2008) concludes that there is no one type of international university.The above stated goal about inclusion and learning of others' experiences and world views will take different forms in different contexts (see Brookes and Becket 2011).Moreover, studies from the Nordic countries demonstrate that a number of languages can be used next to English, depending on the participants' linguistic resources and proficiencies (Ljosland 2008, Söderlundh 2012, see also Lønsmann 2011).English-medium education is nominally in English, in the sense that English is the official course language, but in practice a number of languages can be used in connection with teaching (Söderlundh 2010).Thus, the international university seems to be formed both by global and local influences and flows.

Theory and previous research
The relationship between cultural objects in motion and local take-up of cultural forms is discussed by Pennycook (2007) in a study of the global spread of hip-hop.Pennycook argues that transnational cultural products do not necessarily replace local ones, but are refashioned and given new meanings.Hip-hop artists all over the world use similar patterns of cultural conduct (including certain English expressions) that make hip-hop into a recognizable sub-culture.But wherever it occurs, musicians interpret and negotiate the cultural conduct of hip-hop, so that the music offers potential for local identity (p.96 ff.).Thus, the spread of hip-hop does not represent a plain distribution of cultural forms, but "a layered distribution in which local forces are as important as global ones", as Blommaert (2010: 19) points out in his comment on Pennycook's work.Rather than being a process of homogenisation, hip-hop is characterised by a local take-up of cultural forms, as it is interpreted, negotiated and embedded in local, social relations (Pennycook 2007: 6 ff.).
Transferred to practices in higher education, and more specifically to the strategy of using English as medium of instruction, the theory offers explanations to studies reporting that a number of languages might be used in connection with teaching (see e.g.Söderlundh 2012).The use of English as medium of instruction is a strategy of transnational character in the process of internationalising higher education.In its written form it is a strategy of monolingualism, but when applied in actual educational contexts participants can negotiate the strategy so that other languages also can be used.As Appadurai (1996: 17) remarks: "different societies appropriate the materials of modernity differently" and globalisation is itself "a deeply historical, uneven, and even localizing process".
Similar processes of 'transculturation' are shown by Hult (2012) in an examination of globalisation of English in Sweden as it takes shape in educational policy and practice.On the evidence of national curricular documents and observations of English language educators, Hult (2012: 251) concludes that English has not simply been transported to Sweden as a lingua franca.Rather, the language has been reconfigured for local purposes so that it reflects local, Swedish views and values.In the process, English is partly framed for purposes in Swedish society (such as the Internet, TV, films etc.) and partly framed as a language "through which the local is made global" (p.239); a language "through which people in Sweden project themselves in the world of globalized goods and ideas" (p.240).The duality highlights the fact that globalised linguistic varieties are part of both transnational and local processes.Even though English functions as a worldwide contact language, it can be used and perceived as more than a lingua franca in local and national contexts.When languages circulate around the globe, they become discursively situated in national and local contexts (Blommaert 2010).
A central aspect of linguistic globalisation is the notion of power and inequality (Blommaert 2010).Linguistic resources index social meaning in spaces and situations, and, as Blommaert (p. 33f,194) observes, they change meaning, function and value as soon as they are moved out of a local context.English seems to keep its value in the contexts discussed here, but as Blommaert observes, many speakers of other languages will find their linguistic resources to be of a lower value in globalised contexts.Access to prestigious varieties has always been a question of power, but processes of globalisation have created new and complex markets for linguistic resources.These are at play at the international university as well, and I will come back to the issue of power at the end of my analysis.

The study
The international university appears as an example of a focal point in the world of flows identified by Appadurai (1996).The character of focal point comes from the very process of internationalising higher education, which is apparently imbued with some of the most noticeable characteristics of the era of modernity.Indeed, large-scale mobility, the use of a global contact language and inclusion of others' world views are concerns that apply not only to the domain of higher education, but more broadly to today's globalised world.
With Appadurai's (1996) concept of "world of flows" and Pennycook's (2007) study as a starting point, I proceed by analysing how the transnational strategy of English as a medium of instruction can be constructed locally by students on six courses at a university in Sweden.I focus on norms for language choice and how students can construct norms for when it is appropriate-or not-to use English in actual teaching settings.2By norms I mean shared-explicit and implicitexpectations concerning social and linguistic behavior.The definition is based on a discussion by Gafaranga andTorras (2001: 198, 2002: 10), and that "any action which has been accomplished must be assumed to have been made possible by a specific and discoverable norm" (Gafaranga and Torras 2001: 198).However, norms are not necessarily expressed; rather expected actions are "seen but unnoticed", as Heritage (1984: 116) writes, and they pass without comment.
The data were collected for a larger project3 in 2007 and consist of ethnographic observations of six courses at a Swedish university, interviews with students and staff,4 and video recordings of study situations.Half of the courses were offered within a faculty of science and technology (in the subject areas of engineering and computer science), half in a social sciences faculty (in the area of business studies).The majority of the students were Swedish, although the number of exchange students varied between the courses.In the engineering courses, one in twelve was a student on exchange, whereas among the business students the proportion was between one in three and one in four.None of the exchange students spoke Swedish within or outside the teaching situation, and on no occasion did they show that they understood the language.All teaching staff included in the study could speak Swedish.
The analysis of constructions of norms is based on the video recordings of whole-class teaching situations in the courses studied.From these video recordings (in total 28 hours) I have extracted sequences where students or teachers talk about the medium of interaction and/or act as if its medium is deviant in relation to the particular situation, or in relation to their voiced expectations regarding language choice.For instance, students and teachers might ask for translations into English from other languages and act as if English is the only expected language in the context, or they might argue that a language is inappropriate-or appropriate-to use.In this respect I follow Gafaranga and Torras (2002), who argue that norms for language choice can be witnessed in sequences where participants react towards the medium of interaction as being deviant.Violations of the norm are noticed by participants and make visible to them-and hence to the researcher-a sociocultural context (Gafaranga and Torras 2002: 10).Thus, the construction of norms for language choice is not necessarily a question of decisions between discrete languages, but a discursive construction in which participants point out and orient towards one or more language as the medium for interaction (Pennycook 2007: 136-137, also see Gafaranga 2000).
As a complement to the analysis of how norms can be constructed locally in interaction, I describe what attitudes students and teachers show towards languages and varieties used in the courses and towards students who speak them.The discussion is based on interview data, in which students and staff talk about their views on the linguistic situation and the use of English and Swedish.Nine students and two teachers were interviewed.When discussing general patterns of language choice in the courses I also refer to ethnographic observations.5My focus here, however, is on the video-recorded data and how the transnational strategy of using English as a medium of instruction can be constructed locally by students in six courses at a university in Sweden.

Constructions of norms for language choice
I first (5.1)exemplify how English is constructed as a lingua franca (see e.g.Mauranen 2003, Seidlhofer 2001) among participants with different linguistic backgrounds.I then (5.2) point to some differences between the courses included in the study concerning when, how and with whom the local language, Swedish, can be used.In the last section (5.3)I discuss students' attitudes towards languages and their speakers in the courses studied, given that power relations are an inseparable aspect of linguistic globalisation.The analysis demonstrates the dynamic relation between transnational flows and local take-up of such forms, and illustrates that language choice at an international university is influenced by global factors, yet is still firmly local.

English, a lingua franca in class
My first example of how the strategy of using English as a medium of instruction can be constructed in a local context comes from a course in business studies and a seminar on leadership.The sequence (Example 1) takes place in whole-class teaching and is part of the ordinary teaching.
The teacher has just summarised the seminar and is now talking about the importance of common sense among business leaders.Suddenly, the teacher is interrupted by a student who asks for a translation into Swedish of one of the English words that the teacher is using, namely "prudence".As will be shown in the analysis, the question leads to a short discussion in which the participants make visible their expectations of the linguistic situation.
The teacher answers the student's question by explaining the meaning of the word prudence and points out-in Swedish-that the term is relatively unusual.After a few seconds, an exchange student clears his throat loudly in an unnatural manner (line 15).By doing so, the student draws attention to the teacher's use of Swedish, and simultaneously constructs English as the 'normal' language and Swedish as an unexpected language in the teaching situation.By switching to English and commenting on the language choice, the teacher confirms that English is the language that should be used in the particular context.
Example 1. Prudence (BP 0312v).T: Teacher, S: Swedish-speaking student, X: Exchange student from Canada.The sequence is part of the ordinary teaching and part of an institutional context.The institutional character is not fixed, but depends on how the participants act and whether they orient to tasks and identities associated with the institution or not (see, for example, Drew and Heritage 1992: 22-25).In Example 1, participants exhibit either a teacher's or a student's perspective, and the topic of discussion is related to business studies in the sense that the translation of the world prudence has relevance for the Swedish students' understanding of the subject being taught.Hence, the switch from English to Swedish is made in a sequence of institutional character-a sequence of "on-task" talk-and teacher and students together (re)construct English as the expected language in interactions that are institutional in purpose.
It is also interesting to note the teacher's explanation for talking Swedish.The teacher says-albeit with a humorous twist-that he thought that it was past two o'clock and, therefore, he could speak Swedish.The explanation suggests a division between time in and outside class, which corresponds to a pattern of language choice that was observed in the ethnographic study.In general, English dominates whole-class teaching, but Swedish is used in sequences outside class when Swedes are talking to other Swedish students.The Canadian's laughter and comment (line 20, "now he's starting to speak…") suggests that he has noticed the different norms for language choice in and outside class.
A second example on how students can construct local norms for language choice in the ongoing interaction comes from a group discussion outside class with students in business studies.The students' task is to analyse future problems in the market of clothing design and to write a report to their teacher on the subject.Two of the students are Swedes (called S1 and S2 in Example 2), one is British and one is Greek (named X in the transcript).
Example 2. English (M 0212v).S1: Swedish-speaking student 1, S2: Swedish-speaking student 2, X: Exchange student from Greece.In line 2, one of the Swedes switches from English to Swedish; this language choice is soon re-adjusted by the other Swede through a short, but forceful comment (line 8).The question of "using labour and fabrics" has relevance for all students in the group.But S1 turns herself towards S2 when speaking, and she uses a language that only S2 understands.However, S2's comment ("English") challenges the exclusiveness of S1's utterance and her choice of language.By saying the expected code by its name, S2 makes visible a socio-cultural context in which English is constructed as the normal language for on-task interactions among students from different countries.
In a non-institutional setting, similar instances of directness would perhaps be interpreted as rudeness, or even as a face-threatening act (Drew and Heritage 1992: 24).In Example 2, however, S1 answers the challenge without noticeable annoyance and accepts the suggested language by saying "yeah" before continuing in English.For practical reasons the participants have to agree upon which language(s) to use, for otherwise the group discussion will break down and the task given by the teacher would not be solved.Hence, the institutional setting might explain that S1 accepts the suggested language without noticeable annoyance.
In Examples 1 and 2, the listeners correct the choice of language and construct English as the expected medium for communication among students with different linguistic backgrounds.However, the speaker may also construct norms for language choice.In the video recorded data, this is for example seen in language-related episodes (see Swain and Lapkin 1998: 330) where students ask for help translating non-English words into the English equivalent.The episodes normally follow a three-stage structure in which the student (or the lecturer) in the first stage interrupts herself, points out that she does not know a certain word and then says the word in another language.In the second stage, someone in class gives the missing word and, in the third stage, the first person confirms that she has heard the word and repeats it in English.After the third stage participants resume speaking about the subject discussed before the language question.
The structure is illustrated in Example 3, which shows a transcript of a sequence from one of the courses in business studies.The teacher and the students discuss the company IKEA and whether or not its founder Ingvar Kamprad can be said to personify a good leader.
Example 3. Foundation (BP 0312v).T: Teacher, S: Swedish-speaking student, C: Students in class.In line 8, the Swedish-speaking student initiates a language-related episode by switching to Swedish and saying "I don't know what it's called".The student's meta-comment clearly signals that he is changing languages and that Swedish is not expected in the context (Gafaranga and Torras 2001).Most language-related episodes in the data include switches to Swedish.Exchange students make use of other languages on only two occasions.Firstly, a Spanish student uses her competence in Spanish, asking for help with the English word for "fault".Secondly, a student from French-speaking Canada asks for help with the production of English by switching to French.Similar translations from Swedish, French or Spanish in languagerelated episodes maintain a monolingual English interaction.By switching to another language and simultaneously asking for a translation, the speaker orients as if the other language is an instance of deviance, and he or she constructs English as the only expected language for interaction (Gafaranga 2000: 330).Gafaranga calls the switch a medium-repair, in which a speaker orients towards a linguistic item as an instance of deviance, and simultaneously demonstrates his or her own expectations regarding the choice of language.Hence, language-related episodes show that neither Swedish, nor French or Spanish, is part of the jointly established norm for whole-class interactions on the task; rather, immediate translations into English indicate that these languages are instances of deviance.
In sum, Examples 1-3 show how students construct English as the expected or normal language for on-task discussions among participants with different linguistic backgrounds.The norm is built up by the very use of English and by additional constructions of other languages as instances of deviance.

Restricted use of English
The norm of speaking English does not apply to all situational contexts in class, however.The local language Swedish holds a special position in the studied courses, and sometimes Swedes speak Swedish instead of English in whole-class situations.By doing so they depart from the other consistent norm of speaking a language that all participants can understand, and they challenge the position of English as a lingua franca suitable for all situational contexts.
When, how and with whom Swedish is used varies somewhat between the six courses.In most cases, Swedish is heard in interactions or sequences in class that are preparatory to learning or teaching.The aim and roles at play in these sequences are institutional, but the topic is 'off-task', in that participants are not talking about the actual course subject.For instance, Swedish is used in roll-calls, questions concerning group presentations, queries about forthcoming exams and so on.In these specific contexts Swedish is never oriented to as deviant; rather the use of Swedish in teaching related contexts seems to be part of students' jointly established norm.The link to a specific situational context gives Swedish a function of a contextualization cue (Gumperz 1982), pointing out a certain interaction or sequence as preparatory to the actual teaching and learning.
Indirectly, the use of Swedish in a certain context also functions as a restriction of the use of English as a lingua franca.When speaking Swedish in certain situational contexts, Swedes challenge the functions of English as a lingua franca suitable for all interactional contexts.By speaking Swedish, they highlight the preparatory character of the context, but they also strengthen the construction of English as a language for on-task interaction in class.Thereby, the use of Swedish sheds further light on how norms for language choice can be constructed in the ongoing interaction.English is the expected lingua franca for ontask interactions in class, yet it is not necessarily the expected language for all situations where students from different linguistic backgrounds meet and interact.
The local character of norms for language choice in the courses studied is also evident in the last course where Swedish does not function as a contextualization cue.In this course at the faculty of science and technology, students speak Swedish more often and sometimes they make use of Swedish in interactions that are on-task.They pose questions in Swedish, present their work in Swedish and talk Swedish in discussions with peers (Söderlundh 2010).By doing so, they interpret the linguistic situation in a different way than their fellow students, and they construct norms for Swedish and English slightly differently.Instead of constructing English as a shared lingua franca, Swedes associate English with exchange students, while Swedes-who constitute the majoritycan choose between English or Swedish.Thus, the course demonstrates that norms for language choice might differ between local contexts, and that norms can be constructed by participants on a local course basis.The use of Swedish in on-task contexts in class does not appear in other courses studied; rather it is a firmly local construction. 65.3 Language choice and aspects of power Similar to other implicit or explicit expectations concerning social and linguistic behavior, the construction of norms for language choice is linked to power and inequality.The group of Swedish students sets the agenda and it is, in most cases, their linguistic preferences which determine how norms for language choice can be constructed in the local contexts.Foreign students only treat Swedish as deviant if their Swedish 6 In Söderlundh (2010), I explain these differences with two contextual circumstances: firstly the number of exchange students and, secondly, the approach to teaching.Courses at the faculty of sciences and technology are to a higher degree characterised by a traditional, individual-based approach to teaching.The business study courses, on the other hand, rely largely on groupbased methods, such as seminars and group work.In the latter context, social pressure arises regarding language choice and it is not acceptable to use a language that not all participants understand.peers have already done so (see Examples 1 & 2).They challenge the use of Swedish in whole-class teaching interactions that are on-task, but not in interactions preparatory to these contexts.Although the norms put them at a disadvantage, the exchange students adapt to the expectations of the majority, and it is those expectations which mould interaction.
With this in mind, the Swedish language stands out as a rather powerful resource in the courses.According to the norm, English is the expected or normal lingua franca in on-task interactions in class, but in certain interactional contexts Swedish constitutes a legitimate alternative.For example, one of the teachers exemplifies this when she gives instructions before an upcoming exam in business studies: The ones (pause) the Swedes are of course allowed to write in Swedish and the non-Swedish speaking people are allowed to write in English, or are supposed to write in English.(Teacher in business studies) The status of Swedish is established locally in interaction but, obviously, is also related to the broader context of the university courses.Swedish is the language of the surrounding society, and also the first language of the majority of participants.
The special position of Swedish indicates that language choice rests not only on the basis of mutual understanding.In the interviews, Swedish students say that the use of English impacts on social relations with other Swedes.English is perceived as the exchange students' language and associated with students from countries other than Sweden.Accordingly, Swedes talk English to all exchange students, whether they come from Spain, France or Britain and regardless of their actual linguistic skills.The Swedes, in turn, are not as clearly associated with English, even though their relationship to the language is in fact similar to that of their German and French peers.The Swedes are instead associated with English and Swedish, and this link to two languages rather than one separates the Swedes from other foreign or second language users in the courses.
Accordingly, English functions as a lingua franca in the courses studied, but it is also a language that indexes certain values and views among the Swedes (Hult 2012).As pointed out by Blommaert (2010: 33f, 194), the prestige of certain linguistic resources is linked to spaces and situations, and when linguistic resources are moved out of a local context, they change meaning, function and value.Obviously the special position of Swedish relates to contextual circumstances such as Swedish being the language of the majority and of the surrounding society.In other linguistic markets the value of Swedish would change and its special position would be questioned.
Attitudes towards Swedish are not the only views circulating in the studied courses.Beside the positive attitudes towards Swedish, students show varying attitudes towards varieties of English.The varieties spoken by native speakers are perceived as prestigious and exchange students from Canada, Unites States of America or Australia are asked about English grammar and pronunciation, as exemplified in Example 4.
1 S1: background to this article is that an ongoing 2 process of (1.5) e:: (.) eg-eg[ala 3 S2: [((giggles)) 4 S1: hh I don't really know how to pronounce that 5 (0.5) e: ega-(1.5)egala-(0.5)tarism 6 T: ((points at X)) >why don't you ask< ((points again)) 7 S1: how do you pronounce [the e-word 8 X: [sorry buddy 9 (3) 10 X: oh egalitarian?11 S1: >yeah< the [increasing amount of= 12 C: [((giggles)) 13 =e-egalitarian e: theorists (0.5) that 14 implying tha-that leadership based on 15 inequality is unethical In Example 4, the teacher asks an English-speaking student from Canada how the word "egalitarian" should be pronounced.The question relates to an oral presentation, in which a Swedish-speaking student first uses the word in its written form, but then hesitates when he should say it out loud: "I don't really know how to pronounce that" (line 4).The teacher suggests that the student from Canada should help him to pronounce the word (line 6).Their actions position them as less competent speakers of English.
Attitudes towards different varieties of English impact on the social environment of courses.In interviews, native speakers of English are talked about as particularly useful members in group discussions and other tasks given by the teacher, since they can help non-native speakers of English to solve language problems in texts and oral presentations.Exchange students with other linguistic backgrounds are not associated with similar positive values, and their Englishes are not as valuable as varieties of native English in the local orders of indexicality (Blommaert 2010: 194).Students from France or Spain have to prove that they know the language well enough, but they also have to prove that they are good students who can contribute to the group work.The data thus also exemplify how linguistic resources change value when they are moved out of a local context (Blommaert 2010: 194).In fact, varieties that index native knowledge of English (or almost native knowledge) appear as the only truly, non-negotiable mobile linguistic resources at the international university.

Summary and discussion
The analysis exemplifies how the students in the study can construct local norms for languages in English-medium courses.Rather than talking English all the time, English is oriented to as the expected, 'normal' choice of language when the interaction is institutional in purpose and when the topic for discussion is business studies or technical sciences.As has been shown, however, norms for language choice can be constructed differently in different courses, as negotiations are taking place on a local course basis.Swedish is used in connection with teaching in all courses, but when, how and with whom the language is used varies somewhat between the different courses.In general, Swedish is used in interactions in class that are preparatory to the actual teaching, while oriented to as deviant in discussions that are on-task.The use of Swedish as a contextualization cue strengthens the construction of English as a lingua franca for interactions that are on-task, and restricts the use of English in other situational contexts.
Rather than using English in all study situations, the local norm seems to be based on calculations of other participants' linguistic competence.Students choose the language that seems most fitting to the kinds of roles and aims that the interaction revolves around, but also in relation to their own and other participants' language skills.Yet mutual understanding is not the only basis for language selection, since Swedes sometimes speak Swedish in whole-class teaching sessions where exchange students are present.Particularly, social relations among the Swedish speaking majority influence language choice, and-as has been shown-power relations and aspects of inequality are at play in all studied courses.As pointed out, varieties that index native knowledge of English, or almost native knowledge, appear as the only truly, nonnegotiable mobile linguistic resource at the international university.
In relation to localisation processes, the analysis shows that the transnational strategy of using English as a medium of instruction does not necessarily mean that students speak English all the time.Students can construct local norms for when it is appropriate to use English, Swedish or other languages, and they can reconstruct the transnational strategy of English as a medium of instruction so that it fits local expectations, traditions and ideologies.For instance, English is constructed as a language that first and foremost belongs to the exchange students, while Swedes are associated with English and Swedish.The associations are constructed locally in interaction and illustrate Pennycook's (2007: 94) observation that transnational products and flows can be refashioned and given new meanings in a local setting.The strategy of using English is transnational; still, patterns of language use show that students can construct local norms and (re)create language choice on a local course basis.
The international university appears as a rather stable structure in the world of flows.Its rather stable character comes from the local negotiations of norms as well as local take-ups of English as a medium of instruction.It places the international university among other apparently stable structures, organisations and social forms that Appadurai (2000: 5) identifies in parallel to objects in motion in the world of flows.According to Appadurai, these apparently stable forms function as a structuring force in the world of flows, as they are devices for handling objects in motion.Indeed, with the international university as a rather stable framework, students and staff can (re)construct the transnational strategy of English as a medium of instruction so that it fit local purposes, through direct comments and negotiations (Examples 1 & 2), language-related sequences (Example 3) or use of languages other than English in situational contexts that are off-task.

Conclusion
Like other internationalising strategies in higher education, Englishmedium courses are characterised by interplay between transnational, local and national processes and flows.At the Swedish university studied here, students handle the world of flows by creating and recreating local norms for when, how and with whom English can, and can not, be used.The linguistic environment is first and foremost a local product, even though students and teachers obviously are influenced by the official language choice and by the global function of English as a lingua franca.