Learner–environment adaptations in multiple language learning: casing the ideal multilingual self as a system functioning in context

ABSTRACT Multiple language learning has been largely neglected in L2 motivation research. Recently, complexity principles have been used to model multilingual motivation. In this work, multilingual self-guides are conceptualised as emergent from interactions between the motivation systems of different languages. Motivational systems and their emergent properties are also influenced by the contexts in which acquisition takes place. In this interview-based study in a Swedish secondary school setting, the ideal multilingual self is explored as ‘a system functioning in context’. Focusing on the ways in which multilingual identities and the social contexts of multilingualism co-evolve, analyses show how the school environment shapes and is shaped by emergent identities. The importance of multi-scalar designs is highlighted, and the contribution of motivation research to sustainable multilingual education is discussed.


Introduction
Multilingualism is not a state, but a process. Development is nonlinear and emergent from interactions between linguistic, cognitive and social factors (Herdina & Jessner, 2002). Because language systems are complexly interconnected, separationist approaches are generally ineffective. Indeed, complexity theories may provide the only way in which multilingualism and multilingual acquisition can be adequately conceptualised (de Bot, 2016;de Bot et al., 2007;de Bot & Jaensch, 2015;Herdina & Jessner, 2013;Jessner, 2008).
From a complex dynamic systems theory (CDST) perspective, multilingualism can be understood as an emergent property of interlinked systems that move from one moment of self-organised criticality to another (de Bot, 2008). To investigate development in multilingualism, the researcher faces the task of identifying systems that interact, plotting the relations between them, and conceptualising processes of gain and loss. As a way of hypothesising these relations, and as a precursor to investigations, 'complexity thought modelling' (Larsen-Freeman & Cameron, 2008, p. 41) provides a tool with conceptual and empirical power. In previous work investigating multilingual motivation from a self system perspective (Dörnyei, 2009), complexity thought modelling was used to conceptualise the interrelatedness of motivational systems in the mind of the multilingual learner, and to test hypothesised relationships between the desire to be/become multilingual and intended effort in foreign language (FL) learning (Henry, 2017;Henry & Thorsen, 2018). In the current study, complexity thought modelling provides a means of conceptualising and exploring relationships between motivational systems and the social contexts of multilingualism. Utilising the complexity principles of emergence and co-adaptation, the study seeks to investigate (i) the conditions under which an ideal multilingual self may emerge, and (ii) the ways in which learner-internal systems and contexts of multilingualism co-evolve.

Modelling multilingual motivation: the role of self-guides
Drawing on Markus and Nurius (1986) theory of possible selves and Higgins' (1987) theory of self-discrepancy, Dörnyei's (2009) conceptualisation of L2 motivation involves visualisation of the self as an L2-speaker. In Dörnyei's model, two self-guidesthe ideal L2 self and the ought-to L2 selffunction to generate effort in learning. The centre pieces of self-discrepancy theory, self-guides are 'self-directive standards or acquired guides for being' (Higgins, 1987, p. 321). While the ideal L2 self is a desirable self-image of the languagespeaker/user the person aspires to become, the ought-to L2 self reflects attributes a person believes he/she needs to possess in order to conform to social expectations. Motivation arises when awareness of a discrepancy between an actual and a desired state is disconcerting to a degree that it triggers compensatory action.
In an article exploring multilingual identities, and the possibility that people might develop multilingual self-guidesself-directive standards for being or becoming a person who speaks one or more additional languages -I used complexity thought modelling to generate a number of hypotheses (Henry, 2017). In complexity thought modelling, aspects of a problem 'are described in terms of complex, dynamic systems in order to develop hypotheses for research or plans for action' (Larsen-Freeman & Cameron, 2008, p. 41). In this work, I suggested that the motivational systems relating to a person's different languages would function as components within a multilingual motivational self system, this system being part of an ecology of interconnected systems in the mind of the multilingual language learner (Aronin, 2016;Jessner, 2008). Further, I suggested that when a bilingual/multilingual person learns an additional language, or when a monolingual person learns two or more L2s simultaneously, dynamic interactions between the self-guides of the different motivational systems would take place. Over time, interactions between the ideal Lx self and the ideal Ly self would lead to the emergence of multilingual self guides. These would have system-level effects, creating cohesion and stability in L2 learning behaviours. In situations where ideal selves have a complementary relationship, interactions were hypothesised to lead to the emergence of a self-guide reflecting the person's aspirations to be/become multilingual: an ideal multilingual self. In addition to energy deriving from the desire to become a proficient speaker of the Lx and the Ly (behaviour steered by the respective L2 self-guides), energy would additionally stem from the person's desire to be/become multilingual.
These hypotheses were subsequently tested with secondary students learning English and FLs in Sweden (Henry & Thorsen, 2018). To establish discriminant validity for the ideal multilingual self construct, and to determine whether it would have a unique influence on motivation, a questionnaire containing items measuring the ideal L2 self, the ideal multilingual self, and intended effort for FL learning was administered at two schools where attitudes to FLs and multilingualism were expected to be positive. Structural equation modelling showed the ideal multilingual self and the ideal L2 self constructs were conceptually distinct, and that the influence of the ideal multilingual self on intended effort in FL learning was mediated via the ideal L2 self. In subsequent research, the validity of the ideal multilingual self has also been demonstrated beyond the original Swedish context (Liu, 2020).

Language learning motivation, and the interrelatedness of system and context
As mental representations of potential states, and the cognitive components of hopes, goals, fears and threats, possible selves are individual and personal. At the same time, they are 'distinctly social' (Markus & Nurius, 1986, p. 954). Arising from comparisons with real and imagined others, they are 'made salient by the individual's particular sociocultural and historical context and from the models, images, and symbols provided by the media and by the individual's immediate social experiences' (Markus & Nurius, 1986, p. 954, emphasis added). In CDST, context is a part of the system's complexity. In applying complexity principles to motivational systems, an embodied view of cognition and mental processing is therefore necessitated: A complexity perspective on context in human systems, including language-related systems, insists on the connectedness of the social, physical, and cognitive. Complex systems are open and not separated from context, but interact with contextual factors as they change over time. We therefore need an embodied view of mental activity, including language use and processing, in which mind is seen as developing as part of a physical body constantly in interaction with the physical and sociocultural environment. (Larsen-Freeman & Cameron, 2008, p. 34, emphasis added) The connectedness of cognitive and social systems is similarly highlighted by Duff (2017). Arguing that L2 motivation needs to be re-theorised 'beyond the construct of the individual "self" or "selves"', she makes the case for 'a more holistic, interactive, ecological approach to motivation and other aspects of language learning and multilingualism' (p. 604). As Serafini (2020) makes clear, in contexts of multiple language learning and multilingual motivation, an ecological approach can cast light on 'the complex ways that language learner selves emerge over time through dynamic, reciprocal interactions with context, conceived on both micro-and macro-levels' (p. 134). Importantly, an ecological approach can facilitate systemic understandings of learner-environment adaptation.
Because learner-internal and learner-external systems constantly interact and are mutually constitutive, they should not be 'researched in isolation but rather in a more situated, possibly multi-scalar manner' (Duff, 2017, p. 604). In a multi-scalar ecology of motivation and multilingualism, learner-external systems might be defined at multiple levels. These include the peer group level (people with whom the individual interacts in FL learning), the classroom level (the social settings in which FL learning takes place), the school level (communities in which language teachers work and students learn), the institutional level (language policies and the status of FLs/multilingualism at a particular school), and the public/policy level (educational policy, the position of FLs in the curriculum, and the framing of FL learning and multilingualism in popular and political discourse). While each level has distinguishing characteristics, systems at one level do not exist in isolation from systems at other levels. Rather, each level 'exists only through constant interaction with the others, such that each gives shape to and is shaped by the next' (The Douglas Fir Group, 2016, p. 25).

Purpose
Studying motivation in the contexts in which learning takes place, the researcher is challenged to adopt an 'embodied view of mental activity' (Larsen-Freeman & Cameron, 2008, p. 34). It becomes necessary to work with designs where factors are not 'researched in isolation', but in situated, multi-scalar undertakings (Duff, 2017, p. 604). Continuing the process of complexity thought modelling begun in previous work (Henry, 2017;Henry & Thorsen, 2018), and casing the ideal multilingual self as 'a system functioning in context' (Hiver & Al-Hoorie, 2016;Hiver & Larsen-Freeman, 2020, p. 290), the current study seeks to explore the conditions under which multilingual motivation may emerge, and to describe the ways in which learner-internal systems and contexts of multilingualism co-adapt. To this end, the following research questions were formulated: RQ1: In what conditions might an ideal multilingual self emerge? RQ2: How do system and context co-evolve?

Study and setting
The study was carried out at a secondary school (grades 6-9) enrolling 240 students in a metropolitan area of Sweden. In Sweden, as in many northern European settings, motivation to learn a language other than English (LOTE) is generally low. Compared to English, where students from Sweden are among the highest achievers in Europe, achievement for LOTEs is poor. While roughly 80% of students begin learning a LOTE in grade 6 or in grade 7 (at the CEFR A1 level), motivation tends to wane (Henry, 2012). Despite the enhancement to the grade point average that is generated by studying a LOTEa measure instigated by the Swedish government to promote plurilingualismby grade 9 (CEFR A2 level) the number of students learning an additional foreign language falls to some 70% (Granfeldt et al., 2020).
In the school in focus here, multilingualism was valued and languages were a prioritised part of the subject offering. In accordance with curricula requirements, students learnt English and an FL. For students from families with non-Swedish backgrounds, mother tongue language classes were also provided. However, unlike most other secondary schools in Sweden, students were also given the opportunity to learn an additional FL.
For students in the eighth and ninth grades, language repertoires could, therefore, include in addition to a home language(s), Swedish and Englishtwo of the FLs offered at the school: French, German and Spanish.

Method
In the current study, I conducted focus group interviews with students with experiences of learning English and additional FLs. One interview was carried out with grade eight students, and one with grade nine students. I had been carrying out ethnographic work at the school over a number of weeks, and had got to know the staff and many of the students. These students were selected because I believed they would be interested in engaging with questions involving multilingualism. I also carried out a focus group interview with three of the FL teachers at the school, all with long experience of language teaching. These participants are presented in Table 1. I also interviewed one of the school's principals. Information about the study, confidentiality, and voluntary participation was provided. Consent was obtained from all participants.
Focus group interviews were deemed a suitable method of data collection. The format enables participants to think together about a focal phenomenon, and to react to emerging issues and ideas (Dörnyei, 2007). Carrying out the interviews, my aim was to create an informal environment, to invite discussion, to steerbut not directthe conversation, and to encourage elaboration (Prior, 2018). In the student interviews, participants were invited to describe and reflect on experiences of being a speaker/learner of multiple languages, and of studying at a school where languages had high status. Students were asked whether they identified themselves as multilingual and, if they did, to describe the mental images conjured when thinking about a multilingual identity (see also Henry, 2017). In the teacher interviews, participants were asked about experiences of teaching FLs here and at other schools where they had worked, and whether they felt that the status of languages at the school influenced students' responses to learning. The interview with the principal focused on the school's language policy, and the practical challenges of providing opportunities to learn more than one FL. In all four interviews, participants were asked about the presence of FLs in the school environment, and in social discourse. The interviews were conducted in Swedish. Interview guides are provided in the Appendix. Verbatim transcripts of the interviews were created.
In complexity thought modelling (Larsen-Freeman & Cameron, 2008), the aim is to describe aspects of a problem or phenomenon 'in terms of complex, dynamic systems' (p. 41) and to develop 'understandings of behavior patterns in terms of the systems that produce them' (p. 225). With these objectives in mind, a theory-driven thematic  (Braun & Clarke, 2006). Unlike inductive methods, a theory-driven approach is used to interpret qualitative data with the aim of exploring theoretically generated proposals. Theoretical propositions guide the entire research process: they drive the research questions, inform the data collection, and structure the analyses (Hayes, 1997). Working with the data, the complexity principles of emergence and co-adaptation were used to identify and interpret patterns of meaning. First, the transcripts were read through several times. Next, codes were created and data extracts were assigned to these codes. In a final step, extracts that fitted two a priori generated themes were gathered together. Adapted from Larsen-Freeman and Cameron's (2008) 'steps to guide 'complexity thought modelling' of systems' (pp. 70-71), these were (i) contextual factors forming the conditions under which an ideal multilingual self might emerge, and (ii) ways in which motivational systems and context might adapt to each other.

Results and discussion
The results are presented in sections reflecting these themes. Each section is subdivided into two parts. The first focuses on the social environment, and the second on the school policy.

Prelude
Before proceeding to the results, examples of multilingual identities are presented. 1 In the interviews, students described images connected to a multilingual identity using metaphors that indexed openness, discovery and empathy: I have an image that I am multilingual and I have opened more doors.

(Felicia)
Being able to speak lots of languages, it is a feeling. That you are not like stuck in your little box. You are open to things and go out and discover, so to say.

(Daniel)
You can feel that you are a more social person.
(Gino) It feels good to be able to speak several. But it is not only the feeling of being able to speak several languages. But more the sense of a feeling of having things in common with others.
(Sophie) That students develop multilingual identities is also recognised by the teachers. For many students, they suggest, being able to speak multiple languages has personal importance: Celia: It's an identity, to be multilingual. Joel: Yes. You want to get further. Celia: It's not just about French. Well … they have chosen French, and so they're in French. So anyhow, they chose a language. But it's more like 'I can speak several languages'. It's very important for them. Or many languages. And they often ask 'how many languages do you know'? It's a big thing for them. Ruben: It's probably very important, to be able to speak many languages, yes. Joel: Yes, speaking languages, to be able to speak many languages, it gives cred, it's cool.
Contextual factors forming the conditions under which an ideal multilingual self might emerge The social environment Located centrally in a large metropolitan area, the school was unusual in a number of respects. Unlike a majority of secondary schools, which tend to attract students from the immediate locality, at this school students came from all over the city. As the principal explained, the students and their parents had made intentioned decisions: Coming from homes where parents were engaged in their children's schooling, the principal talked about how students recognised the importance of education, and invested effort in learning: Generally, most of our students are interested in school. They are motivated to learn. And they know that they have to be engaged in order to learn.
Another factor setting this school apart from others are the students' home backgrounds. In relation to social and migration backgrounds, schools in many Swedish cities are highly segregated. Those in the largest metropolitan areas are the most segregated of all (Hansen & Gustafsson, 2016). Here, however, the student population was extremely diverse. Not only did students come from affluent as well as less affluent areas of the city, but as the principal pointed out, the intake was also ethnically heterogeneous: There are many who have another mother tongue. Mostly Arabic, Persian and Tigrinya. They are the majority. But then we also have Mandarin, Hindi, German, English, French, Italian, Urdu, Thai. Almost a third of our students are entitled to mother tongue teaching.
The language teachers also talked about the students' diverse home backgrounds, and how many came from families with cosmopolitan values, and an appreciation of languages: Celia: I also think that it is part of the profile of many of the students in the school. That they come from the types of family who are accustomed to travel, accustomed to being in different parts of the world, and see the value of being able to communicate a little in another language. Because they, often, they come from cultures where they already have many languages. Joel: Yes Celia: I think so. Ruben: Me too.

Joel:
Quite a few come already knowing four languages. 'OK, we'll add one more'. It's a bit like that. Isn't it usually said that the first five languages are the hardest, then it becomes fairly easy? Because you just keep adding, this comes from that, that comes from there, and so on. Even if it's joking somewhat to say that the first five languages are the hardest, I think it's true that if you come here with three or four languages, and you have a mum from one place, a dad from another, and they have lived in a third country, and now they live in Sweden … And of course they know English, and have a neighbor who is Spanish … Continuing, Joel talks about how parents with migration backgrounds can be aware of the importance of languages, and how this can be communicated to their children: The children are second generation, but the parents are often first generation. They have come here and recognized that Persian isn't helpful. And they've been forced to learn two languages, Swedish and English. Or maybe they already had a little English./ … / And I think that there is an understanding of the importance of languages. And without doubt this is mediated to their children, verbally or in some other way.
In addition to an intake comprising students from families with diverse backgrounds, the school differed from mainstream counterparts in a further important way. While teaching was entirely in accordance with the Swedish national curriculum, and teachers were locally recruited, the school was co-located in a building with an English-medium sister-school. At the sister-school, staff were recruited internationally and few spoke Swedish. Here, the FL teachers were all native speakers. Because some of the common spaces of the two schools intersectednotably the building's entrance, staircases, and the school courtyardstudents could become used to encountering (and sometimes participating in) FL conversations. Given the variety of languages circulating in these spaces, the linguistic environment can be understood as of a type that Blommaert (2010, 2013) has described as complex, fluctuating and 'superdiverse'. As the principal explained: We have all of these different languages. Everybody speaks different languages. I often say that here in the building we have the whole world within our walls.
In addition to containing 'models' from which possible selves can be constructed (Markus & Nurius, 1986), a multilingual environment can function to encourage FL use in everyday interactions. This, according to the teachers, meant that learning could be more immediately meaningful: Ruben: Here in the corridors they can chuck out a few words here, talk a little there … yes, it goes quickly. They learn. And it's not just something that's, that they've got do. Some of them become good quickly. Interviewer: Do you think it has an effect, what you've just said, 'I can talk a bit in the corridors', that there are opportunities for use here in this particular school? That hearing languages is an everyday thing? Joel: Exactly, well, you've almost answered the question yourself, but the fact that they hear so many different languages, it becomes natural with a further a language. They hear a lot of Spanish. They hear a lot of French, and some German, and there's nothing strange at all about continuing. Interviewer: Mm.

Celia:
Yes, well, I wouldn't say that they practice it, because I don't think that they always take the opportunities to use the language with people who speak it here at school. But they are aware. And the language is alive. It's not just something you've heard on a computer or in a film. It's for real. The whole time. And sometimes … I've heard them speaking with X [native-speaking teacher at the sister school] and I understood they've talked about this. So they get pretty proud. Because they realize they can make connections between what they learn, and what is real. So the language is for real here. It's alive. It's not just something you learn from a book or on a computer.

School policy
Another important way in which this school differed from counterparts was in providing opportunities to learn a second FL in addition to English. Currently, only a fraction of secondary students in Sweden (around 1.5%) learn a second FL (Statistics Sweden, 2020). Although the low uptake of a second FL can reflect a general lack of interest in learning modern languages (Granfeldt et al., 2020), this may not be the whole story. Together with a shortage of qualified language teachers, timetabling constraints mean that creating space for an additional FL can be logistically challenging. Even in the best intentioned of schools, difficulties involved in timetabling a subject that many students won't take, and which does not generate any additional credits, can mean that additional FLs are not offered. 2 At this school, however, it was a challenge that the leadership was prepared to take on, as the principal explained: We market ourselves as a 'languages' school, that we have a 'languages' profile. It's become like that. / … / In fact we ought to market ourselves even more, as we offer a second foreign language. And there are very few schools who offer this.
The school's semi-official profile as a 'languages' school was something the teachers also spoke about. They described how, on arrival at the school, students could expect to spend time learning languages, and how developing language skills was something they could be proud of and value, as Joel explained: I think also that it is expected of the students that they [learn languages], and that they expect it of each other, and also of themselves. It's part of the school's culture that you do this. From the time that they are in grade six or grade seven, they start saying, 'is it next year that you get to choose your language, and which languages can you learn?' They are already aware of this when they come here to this school, that it's something you do. So I agree with Celia, it is a thing, a culture thing that's part of our school, that we have here. Students know this / … / They feel that it is valued in a particular way, its … they are proud about it, and it's not cool not to learn an additional language. / … / It's part of being here, when you are at our school, and it's … in the brickwork. It's part of the culture. And I agree with Celia that, well, they have English from the start. They are already good at English. So they don't need to, not all of them but most of them, they don't need to put a lot of effort into it. So, why not another language? And then when they start to learn that, why not yet another when it is on offer?
As this teacher indicates, students can come to the school aware of the expectation that they would learn one, or possibly more than one FL in addition to English, and that it is 'not cool not to learn an additional language'.
The sociocultural context of the present constitutes a space in which imagined roles, and future identities are tested and tried out: 'what others are now I could become' (Markus & Nurius, 1986, p. 954). Possible selves are created in interactions in social environments. They are 'negotiated within the framework of one's central social contexts', and 'instantiated in relationships with others' (Oyserman et al., 1995(Oyserman et al., , p. 1217). In the multilingual environment at this school, where FLs are encountered beyond the classroom, and where language learning is 'in the brickwork', conceptions of the self as a speaker of different L2s, and as a person who is multilingual, can be frequently salient. Providing conditions that enable a 'me-as-multilingual' identity to become accessible in working cognition, these contextual factors can be understood as functioning as a part of the multilingual motivation system.
At this school the development of multilingual identities occurred in a climate of coinciding beliefs about the value of languages, in a context where school leaders worked hard to accommodate the learning of additional FLs, and in an institutional environment that was characteristically multilingual and where social interaction could take place across a variety of languages. Not simply a backdrop upon which identities develop, these contextual factors constitute a 'landscape of possibilities' across which identity and motivational systems range (Larsen-Freeman & Cameron, 2008, p. 68).
Although this 'landscape' might have prominent regions of attraction that represent idealised multilingual identities, it can also include regions representing more socially coerced multilingual identities, and reflecting the perception that, in this environment, FL learning is a necessary part of social belonging. As Coetzee-Van Rooy (2014) has suggested, in environments where multilingualism is prevalent, social expectations can become internalised to a degree that people develop ought-to multilingual selves, and where motivation is generated by the perception of needing to conform to prevailing social norms. In such environments, an ought-to multilingual self can direct people to believe 'that if they are not multilingual in this society, they do not 'fit in', because wellintegrated citizens in this society are multilingual' (p. 124, emphasis added). Like the L1 speakers of Southern Sotho and Zulu in Coetzee-Van Rooy's fascinating study of community multilingualism in urban South Africa (see also Coetzee-Van Rooy, 2020), the identity systems of students in the current school can also gravitate to regions of the state space representing an ought-to multilingual self.

Ways in which motivational systems and context might adapt to each other
Having identified the contextual factors that can be understood to be working as part of the system, I move on to consider ways in which context and identity systems co-adapt.
The social environment As we have seen, students at this school could find themselves in an environment where encounters with FLs extended beyond the classroom, and where opportunities to use and experiment with languages can facilitate the development of multilingual identities. When casing a system 'functioning in context' (Hiver & Larsen-Freeman, 2020, p. 290), system and environment are not considered separate entities, but as closely interacting elements.
Since each is open to the other, influences are bidirectional and system and context coadapt in mutually constitutive ways. Because system and context are conceptualised to be in constant interaction, affordances for linguistic and identity development are not 'products' of the environment; rather, they evolve through interactions between individual and environment (Aronin, 2017).
Being part of an environment where languages have meaning beyond school subjects, and where skills have a current as well as a future value, is identified by several students as a motivational affordance. As these grade eight girls explained, the presence of FLs in the everyday environment could trigger a desire to participate in social interactions: Felicia: You get motivated by being able to speak to each other. But it is also that we are in an international school. There are so incredibly many students who come from so incredibly many different places, and so you get interested a little, and we learn these languages in one way or another, and then you can communicate with them in a language other than English. Carmen: The Swedish Education Agency tries to get students to be so motivated, and learn as many languages as possible. Of course, it doesn't work in all schools. Because in most schools, they [students] are just going for a pass. That's just how it is. But like here people want to do better. And then it's also that we have people who speak Spanish, English, all of the other languages here in school. So you can actually benefit from it in your everyday life, instead of just a week in Spain in perhaps ten years' time.
Observing how other people in the school use language, and aspiring to do so themselves, these students' reflections provide a window into the ways in which system and context co-evolve. In an environment where social discourse is not restricted to Swedish or the students' L1s, there is encouragement to participate in TL interactions. Participation in multilingual discourse described by these students illustrates the adaptive relations between system and context. Encouraged to use TLs in social interaction, students are contributing to the development of a milieu that can further stimulate multilingual language use and be inspirational for others: Gino: Well, the school is a … .well, a school with many cultures and many people from different countries, both teachers and students. And that's why, I think, you can become very inspired by others, by the social environment around you.
In the same interview the girls offer further insights into these mutually constitutive influences, and the adaptive relationship between system and context: Sophie: I think it's good, to be able to use the language to speak with other people and be sociable. Interviewer: Mm. And more languages means that you can be more sociable? Sophie: Yes, exactly. Maja: But maybe it's … Maybe you're right. There's a psychology behind the whole thing. That you want to learn more languages so that you can, like, melt in. Interviewer: Mm. Maja: Maybe, yes I think it's probably right. Because, I've never learnt more languages because I can. I mean, 'I can speak so many languages, this one, that one'. Rather it's, well, it's fun when you see other people speaking a particular language, or speaking a number of languages and seeing how they use them. And so you want to do it too. And you think that maybe … well, maybe this is how you get motivated.
While it is easy to see how this language-rich environment can generate the desire to participate in multilingual practices, these complexly bidirectional interaction dynamics also contribute to the development of multilingual identities. Through processes of contagion, the emotional states of individuals and groups co-adapt (Hatfield et al., 1993). Experiencing the desires of others can trigger exchanges of emotional energy, which can lead to a 'synchronized emotional state': Individuals in groups experience rapid, localized feedback about the emotional states of others. This influences their own emotional states, which influences the emotions of others, and so on, creating a collective, synchronized emotional state among a subset of individuals within the population . This emotional contagion may lead to a shared community-identity about how one's feelings about the situation (rather than what one thinks about it) might impact the choices and actions that might need to be taken. (Hazy & Boyatzis, 2015, p. 1) Seen in terms of the emergence of a collective emotional statea desire to engage in multilingual discoursethe identity and motivational systems of individual students and the larger system of the school environment can be understood as co-evolving entities. As systems coevolve, 'changes in the environmental parameters will partially be related to the adaptations that occur within the system' (Allen, 1997, p. 9; see also Byrne & Callaghan, 2014;Larsen-Freeman & Cameron, 2008). Thus, while the environment encourages students to become involved in multilingual discourse, and to develop multilingual identities, it is itself the collective manifestation of students' engagement in multilingualism, and participation in multilingual practices.

School policy
In educational systems, co-adaption takes place between micro-level, learner-internal systems and the larger systems within which they are nested. As Larsen-Freeman and Cameron (2008) explain, 'patterns of behavior at the classroom level interact with patterns of behavior at the institutional level and, beyond that to the socio-political level' (p. 216). As we have seen, this school was unusual in overcoming logistical challenges and offering a second FL. Opportunities to learn a second FL can mean that students who may not have had intentions of learning several languages when arriving at the school, can subsequently become interested in developing multilingual skills, as these grade 9 students explained: While the social environment and the school's policy of creating possibilities for FL learning might trigger a broader interest in languages, influences seem also to travel in the other direction. In talking about the policy of providing opportunities to learn additional FLs, the principal describes how initiatives have also come from students who have become interested in languages, and how students' interests can influence the language offering: For example we have some students who learn Mandarin as a second foreign language. / … / It's a small, small group. And the reason we can do this is because at [the sister school] there are students who learn it, and so we asked them if we could buy some places there. Here we had a very small group in German, and they wanted to learn Mandarin instead. And that's what they do. And then we have an extremely active group of students in our seventh grade who want to begin learning Japanese in the autumn. And they have long list of names. And for us it's fine. As long as we have a group of at least eight students we can offer any language, providing of course we can find a teacher who is prepared to come here and teach Japanese two hours a week. And of course that's not certain, but if we find one, then we'll offer it.
In this example, we can again see how system and context co-evolve. The multilingual environment triggers interest and engagement in FL learning. Motivated to learn additional FLs, the 'extremely active group of students in [the] seventh grade' are spurred to create a petition seeking to have Japanese included as a second FL. In further iterations, the inclusion of languages such as Mandarin and Japanese contributes to the continuing evolution of the multilingual environment.

Conclusion
The research reported in this article represents a further step in the modelling of multilingual motivation. In previous work, the ideal multilingual self was conceptualised as an emergent property of interactions between components of a multilingual language learner's motivational systems, and was investigated using structural equation modelling (Henry, 2017;Henry & Thorsen, 2018). Focusing on interactions between system and context, the current study extends this work. While the emotio-cognitive and identity systems implicated in multilingual motivation can be understood as co-existing in interconnected, learner-internal ecologies, interactions also take place with the open systems that constitute the social contexts of multiple language acquisition. As Hiver and Papi (2020, p. 123) remind us, 'the environment cannot be seen as merely an additional factor among many for consideration when interpreting motivated L2 behavior', but is itself a dimension of the motivational system. In an educational setting where multilingualism was promoted and where TLs were encountered in the social environment, the current study sought to examine the conditions leading to the emergence of an ideal multilingual self, and to explore processes of co-evolution between learner-internal and learner-external systems. Analyses of interview data point to reciprocal adaptations between interpenetrating learner and social systems, and show how the context shapes, and is shaped by students' identities as emerging multilinguals. Casing the ideal multilingual self as 'a system functioning in context' (Hiver & Al-Hoorie, 2016;Hiver & Larsen-Freeman, 2020), and alert to reciprocal interactions between identity systems and external systems representing the social contexts of multilingual acquisition, it is possible to understand how the social environment is a part of the process in which multilingual self-guides emerge.

Limitations, implications, and future research
In the current study, the school is highly atypical. Moreover, the dataset lacks the hallmark heterogeneity that is characteristic of high-quality CDST designs (Hiver & Al-Hoorie, 2020). Despite these limitations, the study offers valuable insights into the interrelatedness of learner-internal and learner-external systems. As Larsen-Freeman and Cameron (2008) explain, complexity thought modelling functions as a 'lead-in' to actual modelling. It provides a way of thinking about a system's complexity, and facilitates the actioning of complexity metaphors. Most importantly, it challenges the researcher 'to imagine the system under different conditions'. This involves the asking of 'what might happen if … ?' questions (p. 71). Reflecting on the current findings, it could prove profitable to ask such questions. What, for example, might happen to multilingual identity development if the student intake changed, if a changed intake necessitated a move to new premises, or if the principal or the language teachers were head-hunted to new jobs? What might happen if the variety of FLs on offer was increased, or if students were able to lean other languages in the school environment, such as Arabic, Persian or Tigrinya? And what might happen if the school embarked on a programme promoting an interrelated vision of languages? It is by posing questions such as these that we can perhaps best understand the mutually adaptive relationships between identity and context in multilingualism.
Although the particular conditions leading to the emergence of multilingual identities at this school are unlikely to be frequently encountered, the lack of such serendipitous circumstances does not mean that 'bridges between second/foreign language at school and multilingualism in real-life communication' cannot be created in other settings (Cenoz & Gorter, 2015, p. 8). In this regard, it is notable that at the other school in which the aforementioned empirical work was conducted (Henry & Thorsen, 2018), and where students were also motivated to learn FLs, undergraduate students from TL-speaking countries systematically carried out internships. Even if interactions with these TL-speakers might not have been as extensive as at the school in focus here, the progressive policy of providing opportunities for TLs to be encountered in real-life communication beyond the classroom can be similarly understood as a contextual condition implicated in the evolution of multilingual identities.
Developing in the participative spaces of classroom interactions, emerging through the collaborative efforts of stakeholders invested in the promotion of multilingual practices, and evolving in educational contexts where institutional policies favour multilingualism, multilingual identities are shaped by the contexts in which learning takes place (Fisher et al., 2018;Haukås, 2016;Henry & Thorsen, 2018). While the identity-nurturing potential of activities that enable students to visualise and explore multilingual futures has been convincingly demonstrated (Busse et al., 2020), the results of the current study highlight the 'socially distributed nature of motivation' (Ushioda, 2017, p. 469). Because research into motivation and multilingualism demands an 'embodied view of mental activity' (Larsen-Freeman & Cameron, 2008, p. 34), and can benefit from multi-scalar designs (Duff, 2017;Serafini, 2020), the study of multilingual motivation needs to be informed by key complexity principles such as emergence and co-evolution. In empirical work, close attention needs to be paid to the mutually constitutive interactions between the different systems implicated in identity development. It is in this way, and by adopting a holistic approach (Cenoz & Gorter, 2011a, 2011b, that the results of motivation research can best contribute to the development of sustainable programmes of multilingual education. How important is it for you to be able to speak other FLs in addition to your mother tongue, Swedish, and English? How important is the second FL [C-språk] that you have chosen (i.e. French, German or Spanish)? Did it make a difference to you which language you chose?
Researchers say this about motivation: Often it comes from a desire to be someone who can speak a particular language: 'In the future I want to be someone who can speak French, or speak German, or speak Spanish'. It can also come from a desire to be someone who can speak several languages. A multilingual person. How is it for you?
If I asked you to imagine yourself as someone who is multilingual in the future, what images come to mind? How would you describe them?