After Study Abroad: The Maintenance of Multilingual Identity Among Anglophone Languages Graduates

ROSAMOND MITCHELL,1 NICOLE TRACY–VENTURA,2 and AMANDA HUENSCH3 1University of Southampton, Modern Languages and Linguistics, Southampton, SO17 1BJ, United Kingdom Email: R.F.Mitchell@soton.ac.uk 2West Virginia University, World Languages, Literatures, and Linguistics, Chitwood Hall 217D, Morgantown, West Virginia, 26506 Email: nicole.tracyventura@mail.wvu.edu 3University of South Florida, Department of World Languages, 4202 E. Fowler Ave, CPR 107, Tampa, FL, 33620, United States Email: huensch@usf.edu

communities-albeit one associated ideologically with power, mobility, and success. Against this global backdrop, many L1 English speakers are more or less contented monolinguals. Nonetheless, a minority of Anglophone young people are attracted to additional language learning and the adoption of a multilingual identity during their formal education. Their motivation for sustained language learning is distinctive, commonly arising from a wish to be 'different,' together with a sense of self-efficacy arising from positive early classroom experiences with languages, rather than from the instrumental motivations that support much learning of L2 English (Busse, 2013;Lanvers, 2017;Thompson, 2017). For such learners who progress to specialist language programs in higher education, study abroad (SA) offers an enrichment experience that can have powerful destabilizing effects on linguistic identity (Block, 2007). Positive outcomes may include the transformation of participants' self-perception from that of language learner to multilingual user, reinforcement of their international orientation, and openness to other cultures or to third spaces (Kinginger, 2009;Oakes, 2013;Plews, 2015;Tullock, 2018). Study abroad participants may also react less positively, developing a heightened sense of national identity, rejecting local norms, such as concerning gendered relations, and/or accepting locals' positioning of them as foreigners and outsiders (Kinginger, 2013). Retrospective surveys of Anglophone SA participants generally show, however, that it is recollected as a life-changing, coming-of-age experience (Coleman & Chafer, 2011;Mulvaney, 2017).
These languages specialists, their evolving linguistic identities, and their eventual career destinations are of particular interest, given frequently expressed policy concerns about the decline of language skills in English-dominant communities, together with perceived societal needs for such skills to promote trade and economic development, security, and intercultural communication (British Academy, 2019; Commission on Language Learning, 2017). This article explores the character of language identity among such Anglophones, once their formal education including an extended SA period has been completed and as they enter the world of work and adult life.

Identity: Poststructuralist Perspectives
Conceptualizations of identity in the applied linguistics literature have been greatly influenced by poststructuralist perspectives that view identity as evolving, dynamic, and performative (Block, 2007;Duff, 2012;Norton, 2014Norton, , 2017Preece, 2016). The structural categories prominent in older accounts of identity (age, gender, social class, ethnicity) are not ignored, but greater scope is acknowledged for individual agency in shaping identity, and its ongoing construction in interaction (McEntee-Altalianis, 2019). For the poststructuralist sociolinguists Bucholtz and Hall, for example, identity is viewed as "the social positioning of the self and others" (2005, p. 586); it is "inherently relational, and will always be partial, produced through contextually situated and ideologically informed configurations of the self and other" (p. 605).
Language itself is highly relevant to such conceptualizations. As Joseph (2016) put it: "Identities are manifested in language as, first, the categories and labels that people attach to themselves and others to signal their belonging; second, as the indexed ways of speaking and behaving through which they perform their belonging; and third, as the interpretations that others make of those indices" (pp. 19-20). These different roles for language are all relevant to the potentially identity-disrupting SA experience, and subsequent evolutions of identity during entry to postgraduate careers.

Language Identities of Instructed Learners
Sociolinguistics has shown how different identities can be expressed through styles, varieties, and linguistic practices within one language as well as through multilingual resources (McEntee-Atalianis, 2019). Our particular concern in this article is with the development of identity in response to the instructed learning of one or more additional languages, and an associated stay abroad, which we will be calling multilingual identity; other related terms in the literature include second language identity (Benson et al., 2013), plurilingual identity (Beacco, 2005), and multilingual subject (Kramsch, 2009). A range of suggestions have been made as to how such instructed learner identities can be defined. As their initial working definition, Benson et al. (2013) proposed that "second language identity refers to any aspect of a person's identity that is connected to their knowledge or use of a second language" (p. 28). Fuller elaborations of this definition are discussed in following sections.
Other researchers introduce varied suggestions as to how the language identity of instructed learners may develop. Many refer to a basic distinction between the identity of L2 learner and that of L2 user. Henry (2017) related different possibilities for identity development to the L2 motivational self system of Dörnyei (2009), equating the identity self-concept with Dörnyei's ideal L2 self. Based on his studies with Swedish high school students learning English and other additional languages (Henry, 2011), Henry (2017 argued that the privileged status of English, in combination with students' perceptions of their own more advanced capabilities in English, may lead to development of what he calls a "contentedly bilingual" self, that is, "someone comfortable and confident in speaking their native language (e.g., Swedish) and the currently dominant global language (English), but not perceiving any additional need or having any particular interest in speaking another language" (Henry, 2017, p. 553). Henry (2017) recognized the possible existence of separate selves associated with each language being learned; in the case of a contentedly bilingual identity, the selves are in competition, and energy invested in learning English detracts from motivation and effort with respect to any third language (L3).
As an alternative, Henry (2017) argued that the different language selves may reinforce each other, leading to emergence of a superordinate ideal multilingual self. This will in turn reinforce motivation for learning a (non-English) L3, as students holding a multilingual identity will be keen to level up their different language proficiencies (Henry, 2017). Henry sees the multilingual self as ultimately transcending "language-specific identities and concerns," in the life of a multilingual person (p. 561). Here, he echoed an earlier argument of Kramsch (2009): "Each of the languages we speak adds its unique dimension to our signifying self that, in its efforts to maintain its autonomy, its continuity and coherence, struggles to become a multilingual subject" (p. 188).

Conceptualizing Language Identities During Study Abroad
In discussions of identity among SA participants, traditional identity categories are foregrounded by some scholars, such as work on national identity (Jackson, 2008;Plews, 2015), or gender (Kinginger, 2008;Trentman, 2015). Others stress the transition from L2 learner identity to that of L2 user (Mas Alcolea, 2017Virkkula & Nikula, 2010). More elaborated proposals have been made by Benson et al. (2013), who identified three possible dimensions of L2 identity that may develop through SA: identity-related L2 proficiency, linguistic self-concept, and L2-related personal competence. By identity-related L2 proficiency, they refer to the development of the sociopragmatic and interactional competence that will allow sojourners to function as a competent person and project a desired identity, for example, as a polite person and serious student, or as a fun young person. By linguistic self-concept, they refer to reflexive identity, including one's sense of self-efficacy and status as an L2 learner and/or L2 user, language affiliations, beliefs, and emotional factors; by L2-related personal competence, they capture sojourners' sense of independence and agency, for example, as a problem-solver and intercultural actor. This framework was developed with reference to L2 English, and Benson et al. applied it to analyze the narratives of Hong Kong student sojourners in a range of Anglophone set-tings. However, the framework is of wider application, and we draw on it in our own analysis.

Development of Linguistic Identity During Study Abroad: Empirical Studies
Concerning the formation of linguistic identity during SA, research with non-Anglophone participants has focused largely on the development of L2 English identity. As mentioned, Benson et al. (2013) developed their model of the L2 self for Hong Kong students undertaking SA in Anglophone settings. In Europe, Kalocsai (2013) has described the development of an Erasmus community of practice among international students in a Hungarian university, and their unifying adoption of English as a lingua franca (ELF). Other European researchers have described similar communities, and their role in shifting participant identities from that of English learner to English user (Dervin, 2013;Kaypak & Ortaçtepe, 2014;Mas Alcolea, 2017Virkkula & Nikula, 2010). Among these studies, that of Virkkula & Nikula (2010) was continued by Räisänen (2016), who tracked the transition of the original SA participants (Finnish trainee engineers in Germany) over an 8-year period, into professional life. This exceptional longitudinal study documented how the participants became confident professional users of Business English as a lingua franca (BELF), managing multicultural teams through English. A contrasting study of Chinese sojourners in Germany by Maeder-Qian (2018) described participants who mostly found it very hard to penetrate local student networks, and whose sense of cultural distance became consolidated over time. While coming from varied regions of China, they mostly associated increasingly with other Chinese sojourners, developing a heightened sense of a shared national identity, and shared linguistic identity expressed through Mandarin. Only a small minority in the group developed clear user identities for either ELF or German.
These Europe-based studies of non-Anglophones are still relatively few in number, small in scale, and mostly limited in focus to L2 English. However they illustrate some complexities of the linguistic experience of SA in multilingual settings, and a range of possible outcomes in terms of linguistic identity.
Regarding Anglophone sojourners, a number of researchers have investigated the impact of SA on identity. Kinginger (2013) reviewed their findings, concerning traditional identity categories of gender, nationality and 'foreigner' status, ethnicity, and linguistic inheritance (Tullock, 2018, presents a similar review). Kinginger agreed with Block (2007) that the SA experience potentially disrupts the cultural habitus of participants. Some react by engaging actively with local norms and practices and embrace a third space (intercultural) identity; this is reflected in their L2 interactional competence (Brown, 2013). Even when positioned as 'foreigners' by locals, participants may view this as an opportunity and exploit it to develop local contacts (Du, 2015). Others, however, cannot adapt to local practices, for example, in gender relations, or in foreigner positioning (Iino, 2006), and may retreat into an enhanced national identity (Kinginger, 2016;Wilkinson, 1998). Both Kinginger (2013) and Tullock (2018) acknowledged that these outcomes may be connected with Anglophone sojourners' typically elite socioeconomic status. They noted the still-limited state of identity research, so that conclusions are generally tentative, including those connecting identity development with the development of multilingualism. Few of the studies they reviewed were longitudinal, and none followed up the development of identity into the post-sojourn stage; for such studies, it is necessary to turn to the general international education literature (see the next two sections).

Identity in Transition to the Labor Market
Before examining research on the long-term impact of SA, it will be helpful to briefly examine contemporary graduates' transition into work, and the development of a so-called "labor market identity" (Tomlinson, 2012) or "graduate identity" (Holmes, 2015). These higher education researchers argue that in the flexible labor market of an economy such as that of the United Kingdom, the transition to work cannot be captured by simplistic models of 'employability,' which see it as the responsibility of the individual graduating student to equip themselves with a particular skill set, matching the requirements of professional employment (Holmes, 2013(Holmes, , 2015Tomlinson, 2010Tomlinson, , 2012. Instead, the transition to work involves a complex negotiation between the structural requirements of the labor market, the preferences of employers (which may relate to social class, ethnicity, and gender as much as to desired employee skill sets), and the exercise of agency by the graduate, who may be influenced by a variety of factors other than career ambition (e.g., emotional ties or geographical preferences, K. Evans, 2007;Finn, 2017). For Holmes (2015), graduate identity is not acquired with the degree certificate; instead it is best viewed as emergent, dynamic, and requiring continuous (re-)warranty by ongoing practices and identifications evolving through interactions in the workplace.
Higher education researchers comment additionally on the phenomenon of graduate underemployment, that is, graduates accepting jobs that have not traditionally required high-level professional skills. Again, they explain this phenomenon with reference to the mass nature of contemporary higher education, and the flexible labor markets of the United Kingdom (and United States), where graduates may move in and out of such lower-level employment prior to (or as an alternative to) settling into more stable professional careers (HESA, 2017;Vigurs et al., 2018). This is especially likely in the case of humanities graduates (Clarke, 2018;Piróg, 2016) and helps to explain the emergent and "processual" nature of contemporary labor market identities (Holmes, 2015).
How these processes affect Anglophone or UK languages specialists in particular, may be glimpsed through research on graduate destinations by the UK Higher Education Statistics Agency (HESA). While the professions of language teacher and interpreter/translator are commonly recommended to languages graduates as most directly using their degree skills (see, e.g., Prospects, 2017), HESA surveys provide a considerably more complex picture.
The most recent longitudinal HESA study surveyed 100,000 UK-domiciled former students who had graduated in 2013, 3.5 years postgraduation (HESA, 2017). Of the languages graduates in the survey, 1 93% were working and/or studying further; median salary was somewhat below that of graduates generally, but actual unemployment was very low. For the languages group, the most popular recorded occupations were "Education" (32%), "Professional scientific and technical" (13%), and "Information and communication" (11%). HESA provides a further breakdown of occupations, as "professional" or "non-professional," and here the languages group underperformed somewhat (though similarly to other humanities graduates). The languages group was judged to have 78% professional occupations, compared with 84% for graduates generally, and 21% non-professional occupations, compared with 15% overall. The most common non-professional category was administrative and secretarial (11% of all occupations, compared with 6% for all graduates).
These findings from general higher education research encourage us to expect that the multilingual dimension of specialist languages graduate identity will continue to evolve in response to labor market entry, as part of a wider ongoing process of identity development. However, surveys such as that of HESA cannot tell us how far the workplace experience is likely to strengthen or weaken that dimension, nor how far languages graduates can exercise agency to seek multilingual workplace (and social) experiences within wider Anglophone society; a more focused and indepth approach is needed.

Study Abroad and Identity: Follow-Up Studies
Some insights into the longer-term impact of SA on the identity of Anglophones are available in the international education literature. A number of international education researchers in the United Kingdom and United States have traced participants some years following their sojourn abroad, to survey perceptions of the impact of SA on the later life course (Alred & Byram, 2006;Coleman & Chafer, 2011;DeGraaf et al., 2013;Dwyer, 2004;Mulvaney, 2017;Nunan, 2006;Potts, 2015). While these studies do not directly address the development of identity post-SA, they report a range of relevant outcomes including impact on careers, on personal development, on international posture, and intercultural orientation. For example, Mohajeri Norris and Gillespie (2009) reported a survey mailed in 2002 to 14,000 graduates of SA programs between 1950 and 1999. They had 3,700 respondents, who claimed significant influences of SA on many aspects of life including career choice: Living and studying in another country engage and affect participants' personal development, worldview, and intellectual and cultural interests, influencing their future decisions. The data demonstrate a sequence of decisions that students make, beginning with the resolution to study abroad, that correlates with the lasting effect of developing a career with a global focus. (Mohajeri Norris & Gillespie, p. 395) DeGraaf et al. (2013) compared SA participants from one particular university with similar alumni who had not sojourned abroad. They found that the SA participants had a somewhat stronger international posture and intercultural engagement than nonparticipants. As in other studies, the SA respondents were almost universally enthusiastic about their experience, reporting a profound influence on personal development. DeGraaf et al. also provided some breakdown by academic major; the strongest influences of SA regarding career choice as well as development of language skills were reported by languages majors (many became teachers).
Researchers reporting long-term post-SA surveys of Anglophones are aware of several limitations to their approach, including various kinds of response bias, the subjectivity involved in selfreport, and the lack of information on whether those choosing to study abroad had any distinctive characteristics pre-departure (e.g., a stronger international posture than their non-SA peers). These issues are discussed, for example, by De-Graaf et al. (2013), who advocated a more systematic approach including longitudinal tracking of particular cohorts. The study reported here sets out to address some of these concerns, on a small scale, by following a well-studied group of undergraduate languages specialists and SA participants into their adult life and early careers. Two research questions are posed: RQ1. What are the career pathways, social networks, and language practices of Anglophone specialist languages graduates, 3 years postgraduation? RQ2. What are the language identities claimed by specialist graduates, and how were these shaped by their educational experience, including an academic year abroad?

Participants
The 33 participants in this study were languages graduates of a UK research-intensive university. Three participants reported a bilingual personal biography (Finnish, Polish, or Welsh plus English). The rest were monolingual speakers of English at home, who had already chosen to become language specialists at high school. At university they had made a further positive choice to study French and/or Spanish to degree level, in combination with cultural or linguistic studies, or other languages (e.g., German), or a humanities or social science subject (e.g., history or business). Other studies of British languages specialists have attributed the decision to study languages at university to intrinsic enjoyment of languages at school, a sense of self-efficacy as language learners, an international posture, and sense of distinctiveness compared with monoglot Anglophones (Busse & Williams, 2010;Oakes, 2013;Stolte, 2015). The participants in this study reported similar motivations. Their BA program lasted 4 years, including a compulsory third year abroad in France, Spain, or Mexico, which, in line with traditional British practice, might be spent studying at a partner university, working as a language assistant in a school or college, or as a workplace intern.

The Languages and Social Networks Abroad Project
As students, the participants had all taken part in a larger (n = 56) longitudinal study tracking their social, personal, and linguistic development over a 21-month period. This was the 2011-2013 Languages and Social Networks Abroad (LANGSNAP) project, (Mitchell, Tracy-Ventura, & McManus, 2017). 2 In the course of LANGSNAP, participants completed a range of language tests, background questionnaires, and interviews on six occasions before, during, and following SA. They generally made substantial linguistic progress (in fluency, accuracy, and lexis, though less so in complexity), and sustained this following the return to academic study at home. Findings also confirmed participants' linguistic self-concept as distinctive "language people" (C. Evans, 1988), and the general evolution of participants' identity over time from L2 learner to L2 user. It appeared that they were aspiring in most cases to a flexible multilingual identity (Henry, 2017), compatible with a mobile future life and career, rather than to an integrationist bilingual identity with prime loyalty to French or Spanish. For example, participants regularly sought opportunities to learn additional central or supercentral languages even when abroad (e.g., German, Italian, or Chinese). Regarding identity-related L2 proficiency, they were very motivated to develop oral fluency, including more informal spoken registers of French or Spanish, and were comfortable about translanguaging practices with other international interlocutors. With some exceptions they were less interested in fully mastering academic registers of French or Spanish; their student identity was somewhat weakened during SA by comparison with their international sojourner identity (though it revived strongly on return to the home university). In most cases they showed little interest in learning regional languages encountered abroad (e.g., Basque or Catalan); the exceptions were those participants who had formed very close personal relations with locals in multilingual regions (e.g., a Mayan-speaking romantic partner in Mexico or Valencian Catalan-speaking work colleagues in Spain). Regarding L2-related personal competence, all reported strong growth in personal independence, and increased awareness of other culture(s), and almost all reported openness to future mobility.

The Follow-Up Study
The LANGSNAP participants graduated from university in 2013. In summer 2016, all participants were invited to contribute to the follow-up study reported here, 3 and 33 of them agreed. In the follow-up study, participants completed the same full set of language assessments as in LANGSNAP, further questionnaires, and an L1 interview. Linguistic findings are reported by Huensch et al. (2019) and Tracy-Ventura, Huensch, & Mitchell (2020). Overall, results show that general L2 proficiency, oral fluency, and lexis achieved at the conclusion of SA had been maintained by a majority, 3 years postgraduation. Quantitative analysis showed that the level of proficiency immediately post-SA was a significant predictor of maintenance of L2 accuracy, while the degree of L2 exposure postgraduation predicted maintenance of L2 fluency (Huensch et al., 2019). In this article, we provide a more detailed qualitative exploration of the participants' life course postgraduation, and their patterns of engagement with languages, to address our two research questions.

Instrumentation
We address these questions primarily through analysis of questionnaire findings, triangulated with analysis of individual interviews conducted with participants.
Language Engagement Questionnaire. This questionnaire was developed for repeated use in the LANGSNAP project and is available in the IRIS repository (Marsden, Mackey, & Plonsky, 2016). Participants are invited to identify all languages that they use regularly, and then complete a separate questionnaire section for each. The questionnaire lists a selection of activities, both face-to-face and online, from home life, academic study, and leisure, and invites respondents to identify how frequently they undertake each activity in a given language, on a 6-point scale. LANGSNAP experience showed that this questionnaire provides a meaningful picture of relative frequency of use, by language and by activity (see Mitchell et al., 2017). A screenshot of the Spanish page is included in the Appendix.
Background Questionnaire. This questionnaire was newly developed from the questionnaire of Mehotcheva (2010). It captured information regarding key aspects of participants' life course postgraduation: further study (including any further language study), jobs, personal relationships (including those involving bi-or multilingual interaction), and travel patterns. The questionnaire also explored participants' current linguistic selfconcept and perceptions of their own language proficiency, through both closed and open questions. Participants' written responses to the open questions were analyzed thematically; the full background questionnaire is available in Supporting Information A.
Interviews. Participants also completed two individual semi-structured interviews with a member of the research team, one in English, and the other in French or Spanish. Before these interviews, the research team members reread relevant LANGSNAP interview material, so as to follow up appropriately on relevant personal information.
The L2 interviews (L2I) served a double purpose, providing a current sample of participants' interactive proficiency, and also providing substantive information on topics similar to the first part of the background questionnaire, that is, activities and relationships since graduation, plans and intentions for the future, and participants' retrospective perspective on their decision to study languages. The L1 reflective interviews (RI) sought fuller information on participants' experiences of travel, personal relationships, and language use patterns, and invited them to reflect on their current proficiency in their various languages and on their language use patterns, as well as on their past SA experience. The interview guides are included in Supporting Information B and C.

Data Analysis
Quantitative data from the two questionnaires was entered into IBM SPSS Statistics 25. All interview data (L2I and RI) were orthographically transcribed. All qualitative data (the interview data plus responses to open questions in the background questionnaire) were imported into the qualitative data analysis package NVivo 12 (Jackson & Bazeley, 2019). As a first step, to facilitate focused reading and rereading of the material, all interview data were autocoded by main interview question. A set of thematic codes was then developed inductively for use in more detailed analysis of the qualitative material, deriving partly from the topics addressed in the questionnaires (e.g., qualitative comments and elaborations on employment histories, on foreign travel, on language learning), and partly from the Benson et al. (2013) identity framework (e.g., comments on priorities regarding L2 proficiency, on the linguistic self-concept, on personal competence deriving from study abroad). So that qualitative findings could be related more easily to the quantitative questionnaire findings, qualitative analysis relevant to RQ1 was also summarized in an Excel spreadsheet providing brief overviews for each participant of employment history, further study, social networking, travel, and language learning. The account of findings for RQ1 given in the next section draws primarily on the questionnaire analysis plus participant overviews; for RQ2, findings derive from the identity-related coding, and are illustrated by representative interview quotations.

Early Career Development
Upon graduation in 2013, it seemed that entry to the labor market was not straightforward for these specialists in languages. Frequent job changes and interruptions to employment were reported. For example, seven participants reported having taught English as a foreign language (TEFL) at some point since graduation, though by 2016, only two were still doing this. A minority were currently making major life choices: to train as a nurse, or as a music teacher, or to emigrate (to Canada, in two cases). Five participants had undertaken further postgraduate study at the master's level (and one of these had proceeded to a PhD in international development). Three had completed a Postgraduate Certificate in Education (the main qualifying route to teaching in the United Kingdom), and several others obtained TEFL qualifications; one took a journalism course. Most were now UK-based, though a striking amount of international travel had been undertaken postgraduation; 17 people reported trips outside Europe, including 6 who reported trips of 6 months or more, and most of the rest had travelled within Europe, with only 3 participants reporting no foreign travel since graduation. Some of this travel involved paid employment or workplace internships (n = 19), though typically this involved short-term TEFL or other relatively low-status, temporary work. Seven were living abroad at the time of the 2016 interviews, in Spain (two), Belgium (two), Australia, France, and Canada; the PhD candidate was studying in her Nordic home country.
By 2016, all participants had undertaken some form of paid employment; at the time of interview, just seven were not currently working, and three of these were doing job-related training. However, only a small minority (n = 7) were committed to graduate-level professional careers involving active regular use of their L2(s) (three languages teachers, one qualified interpreter, one trainee diplomat, one prospective bilingual journalist, and one Barcelona-based business relations manager). 4 A number of others had embarked on long-term graduate-level careers in English journalism, finance or business management, or public administration; this group reported no professional use of their L2 skills. A further group were working in private-sector jobs, where prospects for long-term career development were less clear; these included marketing, office administration, Web site development, and events management. Several from this group reported multilingual practices at work (e.g., managing international events or supporting multilingual Web services for a company or a community). It seems there is a tier of subprofessional multilingual job opportunities available, which these specialist graduates could actively seek out, apart from more formally structured graduate careers with large companies or government.

Social Engagement and Language Practices
Overall, if temporary TEFL jobs are included, a large majority of participants (n = 27) reported in interviews that they had had some form of work experience involving languages. A substantial minority (n = 13) also reported attempts to learn or further develop additional languages. These ranged from participants who needed an additional language for work (e.g., in order to teach Spanish alongside French, to undertake event management in Germany, or with a more distant view of working for a nongovernmental organization), to those who 'picked up' some phrases in a local language while travelling (e.g., Quechua during a Latin American trip). Table 1 presents summarized findings from the Language Engagement Questionnaire (LEQ) concerning current use of French and Spanish (n = 27). Seven participants also reported use of a further language: Italian (2), German (2), Finnish, Polish, and Welsh, the last three being languages of home and family.
The table shows that digital media were the most popular means of accessing and using French and Spanish (internet browsing, social media, texting, instant messaging, listening to mu-sic); around one third of the group reported engaging in these activities at least weekly. Other more extensive types of reading and writing were rarer, though not absent. The subgroup who reported currently using their languages most regularly in face-to-face interaction were those currently living in France, Spain, Belgium, and Canada (Quebec) (n = 6); the three teachers (obviously) also reported regular classroom use of languages.
Information on participants' social networks and related patterns of language use since graduation was drawn from the background questionnaire and from interview analysis. Nineteen people reported a current romantic relationship, including five with international partners, though only one was maintaining an international relationship established during SA in Mexico. Four described their partner as their fiancé(e), though none had yet married. Regarding living arrangements, eight participants were living with their parents; most others were sharing rented accommodation with similar-aged peers, who might include their partner or friends from university, while just one participant owned their home. Like their shifting employment patterns, these living arrangements reflected the lengthy transition to independent adulthood characteristic of contemporary humanities graduates, and the continuing centering of social networks on same-aged peers.
Regarding social networking with L2 speakers, only a small minority had maintained local friendships arising from SA, though rather more had revisited the country of their sojourn. One participant (160) was now engaged to his SA Mexican girlfriend, and 173 had maintained a strong relationship with his sojourn host family and other local friends. Overall, participants were somewhat more likely to report ongoing contacts with international peers met during SA. However, most participants (n = 23) also reported having made new L2-speaking or multilingual contacts since graduation.
Regarding patterns of language use, a large majority (n = 25) reported in the background questionnaire that they were currently using L2 less than during the sojourn abroad. Analyzing both questionnaire and interview data from a longitudinal perspective, Huensch et al. (2019) judged 12 participants to have experienced intensive exposure to L2 over the period since graduation, based on criteria of having lived in an L2-using country for lengthy periods of time, having an L2-speaking partner (currently or in the past), and/or having extensive experience of using L2 at work. Additional to these were the Note. Rows are shaded where activity was also relatively infrequent in English. a Includes one member of the Spanish group, who was now living in Quebec and using French rather than Spanish in daily life. three schoolteachers of French or Spanish, who had rather limited language exposure beyond the classroom. Four participants reported essentially no exposure since graduation; Huensch et al.
(2019) described the remainder as having limited exposure. This group included people who had visited L2-using countries for shorter holidays, but otherwise made little use of L2 apart from occasional social/leisure contacts and uses of digital media (see Table 1).

MULTILINGUAL IDENTITIES CLAIMED AND SHAPED BY PAST EDUCATIONAL EXPERIENCE
Our analysis of the interview and questionnaire data for insights into participants' identity development follows the framework of Benson et al. (2013).

Graduates' Identity-Related L2 Proficiency
Concerning identity-related L2 proficiency, those actively using different languages in the workplace made interview comments 5 describing varied needs for oral fluency: EXCERPT 1 (Participant 125, events manager, L2I) So I found a job with a company which organizes big exhibitions and conferences. My role was to talk to people on the phone and persuade them to attend the exhibitions, which I found very hard at first, because I had to speak French. It was for the French veterinary market. There was a large veterinary exhibition in Paris, and I had to call up vets and say "it's a great idea, you must come to this show" ( …) I didn't like it at all, I was a bit ashamed to be speaking to them. But luckily after three months I was transferred into another department, ( …) and there I worked 50% at the exhibitions in France, and 50% at the exhibitions in London. That was okay, I enjoyed it and I learned a lot, I would say I was working like that for around 18 months.
EXCERPT 2 (Participant 152, business startup, Barcelona, RI) I speak a lot at work, I'm there 12 hours a day and majority is in English. Um [my Spanish is] not as confident as it was before. Well perhaps my spoken is a little bit stronger, I don't do much writing in Spanish apart from some (.) fixed communications that I know off by heart ( …) I mean some things have improved that I can recognize, and some things I've let slip, so something I used to struggle with a lot was subjunctive and I that comes out quite easily for me now. But genders and things like that, because I hear a lot of Italian as well, it's all just gone belly up [laughs]. But ( …) even my English has been affected a bit by everyone else's [laughs] international English so it's just all got mixed up.
As seen in Participant 152's description, mixedlanguage practices were reported for almost all L2-using workplaces; only three participants reported using L2 at work, without any additional qualification. This could also be true in personal relationships, for active multilingual users: EXCERPT 3 (Participant 160, marketing executive, RI) When I'm together with [Mexican fiancée] it sort of all comes back in a couple of days, but yeah it's hard to get that practice in. But I write and text every day in Spanish, and that sort of thing is completely fine (.) Probably not the kind of Spanish you'd want to teach anyone, but it's all grammatical and all fun, and she doesn't correct me at all, so (laughs) that's a good sign. But we have our own sort of language now, and we sort of communicate in our own way, make use of Spanish and then a bit of English, so that's cool. But um yeah anyone else listening in to our conversations wouldn't have a clue what we're talking about (laughs).
As for participants following monolingual English-medium careers, the desired L2 proficiency related essentially to leisure activities, including holidays and media consumption: EXCERPT 4 (Participant 161, events manager, RI) I think (.) I think it's kind of a nice thing to have, to be able to read in another language and to be able to converse when you go abroad.

EXCERPT 5 (Participant 157, university administrator, RI)
So I want to go on holiday to Spain, and I want to be able to. So if we're looking at places to go, (.) I've got a habit of prioritizing Spanish speaking areas um ( …), I feel more confident being able to be in a foreign place because it won't be that foreign. and then also I'll be able to get by ( …) so yeah so we're looking at going to Argentina next year for the honeymoon.
One clear limitation to desired L2 proficiency concerned academic literacy. While several participants had undertaken further career-related studies, none had done so through an L2. The case of Participant 107 offered some insights into this issue. She had been living in Paris since graduation with a French partner and working as an administrator. She now wished to retrain as a nurse and had started studying for nursing examinations in France. But despite her lengthy residence, Participant 107 explained: "I found that it was much too difficult on the level of language, and the fact of writing dissertations in French on medical subjects, it is much too difficult" (L2I). Her solution was to undertake nurse training in England, accompanied by her partner. The easy availability of equivalent training through English meant that Participant 107 was not obliged to meet the challenge of mastering medical French to fulfil her new career goal.

Graduates' L2-Related Personal Competence
The participants reported very generally that their experiences as L2 learners, and specifically their SA experience, had contributed strongly to their sense of personal independence, selfconfidence, and openness to other cultures. This was true regardless of current levels of engagement with multilingual practice: EXCERPT 6 (Participant 124, police support officer, RI) When I was looking for a normal job, I think the fact that I had worked quite independently as a secretary in France, I think that gave me really an edge, because I had done that challenging job in another language on my own. But also I think just being abroad and speaking another language, I feel a lot more confident than I used to be. I'm not shy anymore ( …) and I don't care about making a fool of myself in what I say, which is useful because I speak to the public a lot now. Um so yeah I think it was really beneficial in so many ways.
EXCERPT 7 (Participant 152, business startup, Barcelona, RI) [SA offers] the ability to make you feel more of an international person. (.) You have a lot more awareness of different cultures, certainly you understand a lot more about other countries. And that's not just the Spanish country, you meet a lot of different people, so you learn a lot of random things like how they do something in Italy, how they do something in France. And you become I think a lot more understanding of different things as well.
It was rare for participants to express a reflexive perspective on their own cultural background however. An exception was Participant 178, now teaching Spanish in a multicultural school:

Linguistic Self-Concept
We have seen earlier that by choosing to specialize in foreign language study at university, the LANGSNAP participants were already developing a self-concept as "language people" (C. Evans, 1988) at a young age, and that this was considerably reinforced by the experience of SA. Table 2 presents selected findings from the 2016 background questionnaire, relevant to the maintenance of this overall multilingual identity. The table shows that at least two thirds of the group said they were confident users of French or Spanish, were comfortable with the idea of living abroad, and viewed themselves subjectively as bilingual or multilingual. Over one third had studied at least one further language, and almost all agreed it was important to maintain their language skills.
Responses to open-ended background questions (BQ) provide fuller insight into participants' linguistic self-concept. In support of the view that it is important to maintain one's languages, participants frequently adopted metaphors of return on past effort, sometimes referring explicitly to their specialist degree as a symbol of identity: Only two participants made explicit links between their investment in language learning and their actual or potential career identity, while a few others explicitly rejected such links: EXCERPT 11 (Participant 113, studying for master's in international relations, Belgium, BQ) I spent so much time studying it and I think it could be very useful in my career.
EXCERPT 12 (Participant 111, journalist, BQ) I would like to maintain my French for pleasure, but it is not relevant to my career.
In the RIs however, some participants commented more fully on this link: EXCERPT 13 (Participant 106, language teacher, RI) Actually by the time I left university, and I'm not going to lie, I was a bit disillusioned with it all, and I think I was fed up with studying it for so long. But going back to it, and trying to inspire the little people to enjoy it, has reminded me why [languages] are the best things in the world. Because I get really geeky about it, I love grammar, I think it's really interesting, I think languages are really cool, I think the things that they do to your brain are fantastic, and I love learning about that side of it as well. So I'm getting more and more enthusiastic about learning it more, and kind of seeing myself improve, and seeing my students improve.
EXCERPT 14 (Participant 109, journalist, RI) I think it probably made me stand out from someone who'd maybe just studied English, but it's not a necessary requirement of my job, so it was more like an extra thing, that they were like "oh that's cool." When I say to people "oh I studied French," they're quite impressed, because in my office I'm probably the only one who has a foreign language. (Participant 166, unemployed, RI) As it is important for my lifestyle and how I see myself, losing my Spanish would leave a big hole which I don't think would be easily filled. I enjoy the connection I feel with Spanish-speaking parts of the world and the affinity I feel to Spanish speakers. It has given me confidence to walk into situations and speak [which] I didn't have in the past growing up speaking only English.
As implied in the last quotation, for some participants, the degree of emotional attachment differed between languages that they had studied. In all cases, this preference attached to the SA language: When explaining their self-assessments as bilingual or multilingual (or not), most participants used some reference to interactional competence as a warrant for their claims, in line with their comments on identity-related proficiency: EXCERPT 19 (Participant 122, customer service, BQ) For me, being capable of communicating well in a language means you can speak it, hence I can speak two (or three, as my Spanish is getting there). I cannot claim that my French is as good as my English (chances are it will never be), but I am very capable of expressing myself, and in a generally accurate manner, and so I view myself as bilingual.
EXCERPT 20 (Participant 125, events manager, BQ) I speak French and German too, and I have used both at work. While I am aware of my mistakes and faults, my colleagues who do not speak those languages would consider me fluent, so I would describe myself as multilingual mainly due to their perception of me.
A minority rejected the idea that they were bi-or multilingual despite making not dissimilar comments on their own interactional proficiency: EXCERPT 21 (Participant 129, music teacher, BQ) I can speak French and Italian but don't consider myself to be fluent in a language other than English.
EXCERPT 22 (Participant 152, business startup, BQ) I have never used or thought of myself as multilingual because I only speak English at a native level and Spanish at a relatively high level. I would only describe people who speak an additional two languages at a high level as multilingual. I usually just describe myself as "English and also speaks Spanish, having studied it at university." Thus, overall, whether claiming bi-or multilingual status or not, it seemed that most participants were thinking of themselves positively as bilingual or multilingual users, and no longer learners, and believed they were accepted as such by their interlocutors, even when rejecting claims to nativespeaker proficiency in languages other than English.

DISCUSSION AND CONCLUSION
The framework of Benson et al. (2013) has proved useful in organizing and interpreting our data, and enabled us to sketch the linguistic identity of a group of Anglophone languages graduates as they entered their working lives.
Concerning identity-related L2 proficiency, participants generally prioritized face-to-face interactional competence (now, for uses ranging from close personal relationships, to workplace interactions, to leisure and travel). Many also valued online and social media skills in L2 (email, Facebook, texting, etc.). Some valued reading as a work skill and/or a means of keeping in touch with international sport, fashion, or politics; a few enjoyed cultural products such as novels or films. With few exceptions, L2 academic literacy had marginal value; English was ever present in professional and social life, as a constant alternative choice, with limiting effects for identity-related multilingual proficiency. This Anglophone group presented a clear contrast with the Finnish engineers studied by Räisänen (2016), for example, who were fully committed to using BELF for professional purposes in international workplace set-tings, and whose language skills had developed accordingly.
Concerning L2-related personal and intercultural competence, participants resembled those documented in earlier post-SA surveys (e.g., DeGraaf et al., 2013), continuing to refer explicitly to their SA experience as contributing significantly to their independence, selfconfidence, and problem-solving abilities. Many further viewed SA as having enhanced their employability (providing proof of flexibility, resilience, etc.), and had adopted a clear international posture, evidenced, for example, through interest in mobility or learning new languages. Many also attributed increased intercultural awareness to the SA experience, though few showed clear evidence of an external reflexive perspective on their own national identity.
Concerning their linguistic self-concept, these Anglophone languages specialists were aware both of their privileged status as L1 English speakers and their distinctive identity compared with monolinguals (in line with findings of Lanvers, 2017). They referenced their languages degrees as a continuing part of their graduate identity, even though few had adopted language-related careers. They viewed themselves positively as bilingual or multilingual users, who expected to function flexibly and use mixed-language practices including regular use of English as an international lingua franca. While some identified primarily with one preferred L2, and could in Henry's (2017) terms be described as contentedly bilingual, this identification was not typically attached to any particular location (though there were exceptions). Others, however, were keen to learn and use additional languages, so as to maximize professional and personal mobility, and sustain an international posture; for this group, the label multilingual identity seems apposite. Yet others were settling into primarily monolingual lives and careers, but explicit rejection of L2 identity was absent-even if for this group, the L2 self was operative in very reduced domains (leisure and holidays).
Overall, the degree of continuity with findings from our earlier study (Mitchell et al., 2017) is striking. Participants' priorities concerning identity-related L2 proficiency remained stable from the SA phase onward (most evident in an ongoing preference for oral fluency and continuing disinclination to engage systematically with academic and professional registers). Participants were very explicit about the enduring impact of SA on their developing personal and intercultural competence, reflected in their current personal self-confidence and independence, international orientation, and intercultural openness. Concerning linguistic self-concept, participants who had developed close personal relations with locals during SA (e.g., Participants 102, 160, 173) sustained a high level of commitment to bilingualism, including a strong affiliation to one L2 in particular, and sometimes a continuing relationship with the SA locality. The multilingual identity consolidated during SA was still significant for many other participants. While only a small minority had sought languagecentered professional careers such as teaching, a striking feature was the number of participants who had found ways to sustain their multilingual identity through the workplace, often by taking subprofessional jobs involving some form of multilingual practice. Participants' propensity to add new multilingual individuals to their personal social networks also reflects an ongoing positive orientation to multilingualism, as well as an international orientation. This active identity work was not of course universal, especially among those most integrated into monolingual professional environments. For Anglophone graduates, fulfilling professional lives are available without any expectation of L2 use, and L2 use may be marginalized to leisure and holiday practices, despite unweakened positive perceptions of the impact of SA on personal competence, and positive value of a languages degree.
Overall therefore, the SA experience is of continuing significance for Anglophone languages graduates. It confirms their sense of self as distinctive language people, heightens their sense of self-efficacy as multilingual users, and strengthens their emotional attachment to languages. And on the whole, these identifications remain robust, 3 years postgraduation. However, our findings also suggest some biases in the current SA experience of Anglophone students, which limit the choices open to the multilingual self in the longer term. First, Mitchell et al. (2017) documented a downplaying of student identity among many Anglophones abroad, and a related neglect of academic literacy when abroad. It seems this failure to use the SA opportunity to acquire academic L2 registers may have restricted participants' later choices regarding postgraduate studies, with English-medium programs almost universally selected. In turn, a preference for English-medium postgraduate training is likely to limit multilingual career choices. And finally, the somewhat shallow and 'touristic' engagement with local cultures, common during SA, is re-flected in the continuing enthusiasm for touristic travel, and rather limited intercultural perspectives found among graduates. An SA experience that included more challenging academic studies, together with systematic training in ethnographic interpretations of culture and opportunities for critical reflection on the self, the home culture, and on English language practices could provide a more solid foundation for the development and sustained maintenance of multilingual identity among this distinctive group. These suggestions from a longer-term perspective are in line with those of many other scholars who have focused more immediately on the SA experience, such as Jackson (2014), Kinginger (2011), and Roberts et al. (2001). Further support could come from the adoption more generally in the higher education languages curriculum, of a sustained focus on advanced literacy, including systematic introduction to a range of professional genres, as advocated by, for example, Byrnes, Maxim, and Norris (2010) and Ryshina-Pankova and Byrnes (2017).
An adjustment to the languages curriculum by itself will of course not necessarily stabilize and enhance multilingual graduate identity, which develops over time through ongoing interaction between the changing conditions of the labor market and the agency of the individual young adult (Holmes, 2015;Tomlinson, 2010). To understand better how multilingual competence and identity are sustained long term, more detailed research is clearly needed into longitudinal interactions between individual graduate agency and workplace communities of practice. We need to better understand the apparent willingness of languages graduates to undertake short-term subprofessional jobs involving multilingual practice, alongside their apparent reluctance to enter language-centered professional careers such as languages teaching, as well as the opportunities or obstacles within the workplace itself for maintaining and developing different domains of L2 proficiency. We also need to better understand how workplace demands interact with engagement in multilingual social networks and leisure activities. Considerable research attention has been paid to SA over time, but the insights gained concerning identity development in disruptive new settings need to be applied to a much wider range of contexts, over longer time cycles, if we are to meet the expectations of policymakers in creating and sustaining the multilingual expertise required to overcome a current "language deficit" in Anglophone society (Commission on Language Learning, 2017, p. 5). NOTES 1 The HESA categories combine foreign language specialists with English language and literature specialists. 2 The LANGSNAP project was funded from 2011-2013 by the UK Economic and Social Research Council (Grant no. RES-062-23-2996).
3 The follow-up study was conducted with a Language Learning Small Grant and funding from the University of South Florida. 4 The small numbers of graduates entering specialist languages careers is not due to lack of opportunity, for example, there is currently a shortage of language teachers in UK schools. 5 Quotes from the L2 interviews have been translated into English.