Language Intermediaries and Local Agency: Peacebuilding, Translation/Interpreting and Political Disempowerment in ‘Mature’ Post-Dayton Bosnia–Herzegovina

The peace negotiations ending the 1992–95 Bosnian war established a constitutional system of ethnic power-sharing giving international peacebuilding a significant role in post-conflict Bosnia’s governance and economy. This indefinite mission depended – like any other intervention where foreigners work across linguistic boundaries – on interlinguistic mediation by locally-recruited translators/interpreters, an aspect of knowledge production that even current research into peacebuilding’s micropolitics often neglects. On an individual level, their frequently-overlooked agency was integral to peacebuilding practice. Yet theorizing their agency must also acknowledge the macrosocial level, where critics argue the post-war constitution has stripped Bosnians of political agency, foreclosing political participation as anything but ethnic subjects corresponding to three institutionalized ethnic identities. The entrenched, growing disconnect between political elites and the public, expressed through social protest in 2014, foregrounds agency and dis/empowerment in Bosnian society more sharply than pre-2014 research accounted for, revealing further interlinkages between international peacebuilding and domestic political contestation.

agencies have depended on spoken and written language mediation by thousands of locally-recruited translators and interpreters, and much smaller numbers of their own staff (Kelly & Baker, 2013;Buckingham, 2016). 1 These intermediaries' experiences span more than two decades of post-conflict, post-socialist, Bosnian society. In the first years after the Dayton Peace Agreement was signed in December 1995, greater hopes for democratic reconstruction and fairer distribution of socioeconomic resources were at least possible, even if remote. By 2014, the year of the 'plenum' protests against the 'Dayton' constitution and its ethnopolitical elites, it should have been clear to foreign interveners as well as Bosnians that political and economic power still lay with those who had accumulated it in war. The very length of time this system had lasted, indeed, was now shaping Bosnians' sense of disempowerment (Jansen, et al. 2017). This widespread feeling of disempowerment that continues to constitute everyday life in BiH for most Bosnians is a sobering structural counterpoint to the emphasis that Translation and Interpreting Studies (T&IS) now places on interpreters' and other language intermediaries' agency (see Inghilleri, 2005a).
Against this background, this article reconsiders evidence from narratives of 31 locally-recruited interpreters and 20 foreign military and civilian language intermediaries I interviewed in 2009-10 during a project on the language policies and practices of peacekeeping in BiH (see Kelly & Baker, 2013), in light of fresh conceptual frameworks from critical peacebuilding studies and from the social anthropology of postsocialist, post-conflict BiH. 2 These interviews had taken place at a potentially liminal moment when tentative steps towards 'non-ethnic' or 'post-ethnic' political alternatives in BiH seemed to be being made (Touquet, 2011), and when EU and USA negotiators at North Atlantic Treaty Organization (NATO) headquarters in Sarajevo were brokering constitutional reform talks (the so-called 'Butmir Process') that yet again 'focused […] only on political leaders of major ethnic groups' (Dzǐhić& Wieser, 2011: 1817. The prospects of reform, stagnation, and further destabilization were simultaneously in the airand remained so as post-Dayton BiH 'matured' under continued international tutelage, while Bosnians' hopes of improvement continued to wane.

Peacebuilding, translation/interpreting, and agency
Studies of peacebuilding and of translation/interpreting are both concerned with agency, but each field approaches it on different scales. In critical peacebuilding studies, one central concern is why international peacebuilding in BiH and elsewhere has so often failed to engage with or even recognize the 'critical and resistant agency' 1 For background on the war's origins and course, see Baker, 2015b. 2 These interviews were semi-structured, with participants recruited through snowball sampling combined with advertisements in publications, mailing-lists and online groups read by ex-peacekeepers or ex-interpreters (Kelly & Baker, 2013: 12-15). My positionality as a white British woman interviewing both 'international' and 'local' interviewees is discussed further in previously published work (Baker, 2012(Baker, , 2015a.

LANGUAGE INTERMEDIARIES AND LOCAL AGENCY
of everyday peacemaking that local initiatives, in and beyond what interveners regard as formal civil society, can exercise (Kappler & Richmond, 2011). Some research has made structural critiques of top-down intervention models that constrain 'local agency' (Pugh, 2011), and other studies seek 'local spaces of agency' outside international peacebuilding's formal mechanisms (Kappler, 2014: 125). T&IS, meanwhile, had well-established interest in language intermediaries' agency by the 2000s, seeking to make 'translators and interpreters more visible as social actors' (Inghilleri, 2005b: 142).
Whether translation scholars understand agency through 'sociology', 'culture' or 'cognition' (Chesterman, 2009: 13), translators' 'positioning' (Baker, 2010: 197), or social 'habitus' (Inghilleri, 2005a), those working from this perspective would agree that language intermediaries' social situatedness provides much of the context for their choices about what and how to translate. The problem of how much agency they can exercise thus extends well beyond how much creativity a task allows, into the configurations of power and ideology around them, where 'individual and social dimensions' overlap (Palumbo, 2009: 9). This level of agency is much more granular than how agency is theorized in all but the most micropolitical and everyday peacebuilding research.
T&IS theorizations of agency have also, against the background of post-9/11 warfare, enabled scholars to examine language intermediaries' complex roles in narrating and producing knowledge about war, conflict and security. Whenever wars are fought or missions conducted across linguistic boundaries, militaries and states depend on translation and interpreting for their understandings (as accurate, flawed or distorted as these might be) of the conflict and parties' history and culture (Inghilleri & Harding, 2010;Rafael, 2016). Translators and interpreters are therefore, as Baker (2006) influentially argued, heavily implicated in producing and disseminating narratives of conflict, where their social positionality is never fully set aside. It is in conflict situations where agency, within military command/discipline structures and amid the intensity of war, becomes most entangled with translator ethics, especially when linguists take direct part in interrogating and torturing prisoners as at Abu Ghraib (Inghilleri, 2009). T&IS scholars have thus converged with war and security scholars around an interdisciplinary agenda for studying translation/interpreting in conflict (Footitt, 2012).
Today, languages and conflict research covers language training, propaganda, intelligence-gathering, multilingual armies' and coalitions' interoperability, and troops' intercultural encounters with civiliansand, since war's entanglements extend beyond the battlefield into spheres of society normally viewed as civilian (Sylvester, 2013), perhaps even to translation/interpreting in war media, international war crimes tribunals, and asylum systems handling conflict refugees. Importantly, this body of work has recognized language intermediaries as a distinct type of conflict participant, with many examples from twentieth and twenty-first century wars (see Footitt, 2010;Kelly, et al. 2019). Among these are the Yugoslav wars, in which locally-recruited interpreters' complex ethical responsibilities and divided loyalties had already started being studied (Stahuljak, 2000;Dragovic-Drouet, 2007) before Hilary Footitt and Michael Kelly's research on 'Languages at War' (see Footitt & Kelly, 2012) incorporated the case study of peacekeeping in BiH for which I worked. Across these diverse conflicts, researchers find that interpreters 'become de facto players in a conflict which they may not choose but which they sustain both morally and instrumentally, operating 'simultaneously as free agents and embodied conduits for the political and military institutions they agree to serve' (Inghilleri, 2010: 175).
Much more languages and conflict research concerns armed conflict and its direct aftermath, however, than international peacebuilding after war's immediate end, even less the protracted, continually-extended peacebuilding experienced by contemporary BiH. Today, BiH is still being 'peacebuilt' in the sense that international organizations and institutions are operating there to ostensibly prevent conflict recurring, while few members of the public believe they have any agency either to affect the dominant ethnicized frame of local politics or determine BiH's socioeconomic course, and political and socio-economic disempowerment in 'mature' post-Dayton BiH are even more visible now than in 2009-10.

Agency and the everyday practice of translation/interpreting in peacebuilding
In 2009-10, when I was interviewing, an interpreter who had begun working for foreign forces when the United Nations Protection Force (UNPROFOR) entered their hometown then held successive contracts with different military contingents might have spent 17-18 years in work they had never expected to do before Yugoslavia collapsed and war broke out, and in a profession that, outside the NATO headquarters structure (see Jones & Askew, 2014), military employers often did not even see as a professionalthough the most 'continuous' aspect of their employment would have been constant awareness of how quickly it could end if forces withdrew. More commonly, individuals' work histories included a few years of UN/ NATO interpreting (especially in 1995-99 when foreign troop numbers were largest) between contracts with other foreign agencies: there, they might work directly as translators/interpreters, as administrators, or as project officers who still acted as linguistic-cultural brokers between foreign officials and local institutions and society.
Since few interpreters for foreign forces had been professional linguists, that profession's codes of ethics, impartiality and neutrality were not necessarily part of how they perceived their role. Members of better-prepared armed forces received 'use of interpreters' instruction in pre-deployment training that provided a rough equivalent (Baker, 2012: 151-152), teaching them interpreters should communicate military speakers' content and register as accurately as possible to local interlocutors, and relay local speakers' words likewise: interpreters' military supervisors passed on this expectation. Sometimes, however, this conflicted with interpreters' own conclusions that a different approach would better achieve peace. One interviewee, employed by British peacekeepers in 1994-95 (and later by UN/EU police taskforces), explained this through a common situation UNPROFOR interpreters near UN 'Safe Area' enclaves encountered during the war: interpreting at meetings between peacekeepers and Army of Republika Srpska (VRS) troops who had returned artillery and equipment to demilitarized zones. Such encroachments were frequent as the VRS tested how far it could push UNPROFOR before the UN would fulfil its threats of air strikes from NATO aircraft.
All the forms of social practice that Mac Ginty (2014: 556) regards as practices of 'everyday peace' in divided societies -'avoidance', 'ambiguity', 'ritualized politeness', 'telling' (inferring interlocutors' identities from social cues and taking appropriate action) and 'blame deferring' (blaming trouble on outsiders within one's own group rather than escalating a conflict)were available to interpreters in negotiating peacekeeping encounters, as were emotional labour strategies with which interpreters could affect the encounter. If for instance a peacekeeper made a forceful, angry performance to a VRS commander, but an interpreter used their own situated local and cultural knowledge to judge that this would be more likely to provoke the commander than persuade him, the interpreter could 'soften some hard words' and redact profanities, diluting the speakers' linguistic performances but keeping the meeting intact. The interpreter offering this explanation called it 'my contribution to maintaining the peace'. 3 Interpreters' dilemmas in volatile wartime settings, or immediately after Dayton when peacekeepers still faced hostile encounters with armed local troops, exemplify the observation that, throughout conflicts, 'ethical decisions are required of interpreters and translators that extend beyond the translation of a spoken utterance or a written text' (Inghilleri & Harding, 2010: 166). 4 Agency here meant navigating at least three different ethical codes and understandings of role: foreign forces' (and, if different, interpreters' employers' 5 ) expectations that interpreters would faithfully relay content and register; interpreters' own understandings of how best to achieve peace at the micropolitical levels to which meetings pertained; and the expectation of most local military interlocutors, and many (not all) civilians, that interpreters would and should act in the ethnopolitical interests of the group their name and background attached them to (Baker, 2015a). Interpreters as well as peacekeepers thus participated in the 'embodied performance[s]' of security (Higate & Henry, 2009: 17) that constitute peacekeeping and post-conflict peacebuilding.
Interpreters' social situatedness frequently gave these performances higher stakes for local intermediaries than foreign peacekeepers. Bosniak, Serb or Croat interpreters conveying adverse news to commanders from forces representing that ethnonational identity, or to civilians who perceived they shared an ethnic identity, would often face interpersonal pressure to side with their nation not foreign troops. In their hometowns, they were also conscious that interlocutors who held local military or civilian power stemming from the war and/or were connected to organized crime would know or could discover who they were. One interpreter with a Serb background who had moved to Banja Luka, the capital of Republika Srpska (RS), from Croatia in autumn 1991 (at the height of the conflict there) described feeling uncomfortable during anti-corruption and anti-trafficking operations because 'many people […] knew me from the town'. Conscious that many townspeople viewed NATO as occupiersespecially during and after NATO's air-strikes against Serbia and Montenegro during the 1999 Kosovo Warhe understood he could be seen as 'a traitor' who had 'gone to work for the local occupiers' because of the job he had taken from economic necessity, and he alluded to having received threats after incidents such as being seen in public with troops protecting Bosniaks from Serb rioters when the Ferhadija mosque cornerstone (destroyed by RS authorities in May 1993) was re-laid in May 2001. 6 The distancing strategy this interpreter, and many others, employed in such instances was to displace the unwelcome outcome on to foreign troops by explaining that 'I like the cause that we are fighting for and I don't care about these Brits […] we can't throw them out so I'm just working for them and nothing else'. He had used it with his 'own' side and even when (using another common threat reduction strategy among interpreters) he had allowed a Croat Defence Council (HVO) commander to think he was a Croat by stating a name and hometown implying Croat ethnic origin. 7 Such performances were most necessary in wartime and in the first few, most unstable post-Dayton years. While they were less likely to be necessary even in military peacebuilding encounters by the mid-2000s (when the implementation of disarmament was over, and NATO's main peacebuilding tasks in BiH were civil-military cooperation and defence reform) they remained among the repertoire of strategies that interpreters, and Bosnians generally, had for avoiding violence and aggression and negotiating 'everyday' peace.
Understanding language intermediaries as active participants in peace negotiations also throws new light on peacebuilding's embodied gender dynamics. Feminist peace scholars have often observed Bosnian women were absent from the Dayton negotiations and that the constitutional settlement marginalized women's political agency, especially when compared to women's political participation in socialist Yugoslavia (Cockburn, 2013: 27 Ibid.

LANGUAGE INTERMEDIARIES AND LOCAL AGENCY
interpreter into the beginning of a diplomatic career) and his daughter Sabina Ber-berović (McLeod, 2016: 39). Indeed, for the famed 'invisibility' of the translator (Venuti, 2008) not to be compounded by the invisibility of women, feminist T&IS as well as feminist peace and conflict studies deserves to pay much more attention to the sheer amount of affective, and often aesthetic, labour that patriarchal norms push female language intermediaries to invest in creating conducive atmospheres for negotiation. Among foreign peacekeepers in Bosnia, for instance, it was tacit knowledge that bringing young, attractive female interpreters could smooth over difficult liaison meetings with local military commanders (Kelly & Baker, 2013: 60)likely exposing them to degrees of sexual harassment they would have been expected to withstand.
As regarded peace operations' language policy, meanwhile, locally-recruited language intermediaries had some limited agency over how foreign forces negotiated language recognition politics in BiH. 8 All organizations issuing translated letters and documents had to decide whether to communicate in ways performing demonstrable differences between three languages (Bosnian, Croatian, Serbian), as post-war nationalist parties demanded, or to write in only one language variety and expect all recipients to accept the languages were still mutually intelligible. The grammatical and lexical features that mark sentences as Bosnian, Croatian or Serbian form part of Bosnians' everyday knowledge of ethnopolitics, allowing writers/speakers to adapt either to a certain variety or a more ambiguous, inclusive standard by choosing between them. Foreign forces' solutions, especially in writing where differences were more obvious than speech, often came down to efficiency and practicality: for instance, letters to Serbs could often not be typed in Cyrillic if staff had no Cyrillic typewriters or fonts (Jones & Askew, 2014: 57). However, they also conveyed symbolic messages, intentionally or not, about the force's position on language recognition. Translators mediating for multiple recipients would often choose to minimize ethnicised linguistic features by avoiding constructions marked as ethnically specific, while for single recipients they could draw inferences from personal names and tailor text if desired. While force commanders could decide how far they wished to accommodate or resist language recognition demands, locally-recruited linguists produced the documents that local recipients understood as evidence of the force's language policy.
Beyond the reliance that interveners placed on locally-recruited interpreters to translate all but the most sensitive information (reserved for the much smaller number of military linguists and security-cleared civilian contractors), informal interactions between interveners and interpreters also heavily influenced what knowledge interveners gained about communities. Peacekeepers received finergrained local cultural and historical knowledge while travelling with interpreters than they did from pre-deployment briefings, handbooks and training (Kelly & Baker, 2013: 126-127), which concentrated heavily on local armed forces (in NATO parlance 'former warring factions') and their structure and strength. It was from interpreters that troops were most likely to learn how a building on their patrol route had burned down, or how a town had experienced events such as the May 1995 Kapija massacre in Tuzla (when VRS shell-fire killed 71 people celebrating Youth Day), in which many interpreters working for US forces headquartered in Tuzla after Dayton had lost friends. These interactions were potential conduits for peacekeepers to encounter knowledge that was not filtered either through ethnopolitical lobbying or the tropes of their home countries' mediaat least if they could unlearn frames of knowledge they were already accustomed to.
The extent to which peace operations' knowledge production is mediated through locally-recruited interpreters in fact suggests language intermediaries are far more central to this process than even the most granular study of knowledge production in peacebuilding, Séverine Autesserre's Peaceland (Autesserre, 2014), makes them appear. Autesserre's fieldwork-driven approach, hailed as groundbreaking in peacebuilding studies (see, e.g. Cunliffe, 2015), does identify language competence as important, recognizing for instance that the lack of local language competence among most foreign staff is a key reason why international agencies 'discount […] local knowledge' and thus fail to understand the conditions for peace in that society well enough to fulfil their aims (Autesserre, 2014: 80-81). Meanwhile, she argues, intervention agencies' dependence on written reports in English blocks knowledge which it is more difficult to linguistically and culturally translate from reaching managers and donors who make intervention policy at higher levels. Autesserre's model is more sensitive to the micropolitics of language and translation than most peacebuilding research but still takes a curiously instrumental view of locally-recruited language intermediaries themselves.
Peaceland's few references to local interpreters cast them either as low-skilled or as risk factors, more likely to be obstacles to peace than potential agents of it. Recommending that field teams should include local staff in more substantive roles than 'just a driver or translator' (Autesserre, 2014: 262), for instance, advocates for a welcome upskilling of local employeesbut deskills language mediation itself. The 'numerous' incidents Autesserre (2014: 118) observed during fieldwork where 'interpreters routinely make translation mistakes or omit vital information', meanwhile, show that language support is not a 'fail-safe' mechanism for alleviating deficiencies in local cultural knowledge which (she rightly argues) prevent interveners understanding the conditions and meanings of peace. These risks in language mediation undoubtedly exist; but so too do interpreters' own strategies for avoiding them, which go unnoticed unless an interlocutor knows both languages and is closely following the conversation.
Locally-recruited language intermediaries' work in mediating how peacebuilders and other foreign interveners understand local ethnopoliticsand even mediating what stances intervention agencies represent themselves as taking towards local ethnicised language politics (as with their negotiation of symbolic functions of language and word choice in post-war BiH)might from a T&IS perspective constitute a largely unacknowledged dimension of local agency in peacebuilding at the micropolitical level. Following T&IS in tracing local interpreters' positionality, moreover, also exemplifies Stefanie Kappler's observation (also drawn from researching peacebuilding in BiH) that in international intervention 'the local' and 'the international' are processually constructed, rather than fixed and pre-determined, categories, and individuals could change position in them over time (Kappler, 2015: 876).
Besides the fact that 'localness' and 'internationalness' were a spectrum not a binary in BiH (several interpreters I interviewed had for instance been born in neighbouring Croatia or Serbia, and moved to BiH either as wartime refugees or, later, to take up well-paid but short-term jobs with NATO), how far individual language intermediaries belonged to 'international' and/or 'local' milieus was and is a matter of social practice, embodied through characteristics including what language(s) they chose to speak socially and how far they chose or could afford to participate in internationals' leisure practices (Koutková, 2017: 110). These forms of 'delocalisation' or '(re-)localisation' (how far someone might seek to be framed as (not) part of 'the local') represent, for Kappler (2015: 878), a 'process of positioning oneself in the wider peacebuilding landscape', and also constituted performances of socio-political identity within (post-)Yugoslav codes of cultural consumption. Certain factors that individuals themselves could not alter easily nevertheless imposed structural limits on this fluidity: a British or German intervener could resettle immediately in western Europe after leaving BiH, while local employees would need expensive visas to settle there (though their posts and relatively high salaries in local terms might help accumulate the necessary capital and contacts to obtain one). Even though 'local'/'international' boundaries could be porous in the everyday politics of intervention, there were still constraints over how 'international' locally-recruited workers could become. The 'international'/'local' divide, however, by no means circumscribed the whole range of identity categories that interpreters negotiated: different ethnopolitical identity categories, and attempts to articulate non-ethnic alternatives, were highly significant in Bosnian everyday practice, but all equally 'local'.
If interpreters engaged in everyday social practices of peace, and if they were deeply implicated in peacebuilding's politics of knowledge production, did they also represent sources of local agency as peacebuilding studies would understand itthat is, the capacity to influence what 'peace' might mean locally and/or internationally and how 'peace' could be achieved? From their perspectives in 2009-10, those interviewees who did perceive interpreters as having had some influence over everyday peacebuilding offered two main suggestions about what that influence was. Firstly, some (not all) Sarajevo-based interpreters considered that interpreters had been able to contribute to reconstructing BiH by assisting intervention agencies to fulfil projects such as anti-corruption initiatives or material rebuilding, or in wartime simply by interpreting at meetings that brought opposing sides closer to making peace. This was one possible 'activist' stance through which interpreters might narrate their past or present self, though perhaps more often their past.
A different form of influence, narrated by some interviewees who described themselves as Serbs (especially those living in the RS who had worked with British troops) while explaining how they had become interpreters, was the opportunity to present an alternative image of Serbs to the preconceptions they expected peacekeepers to have formed from Western media. This was variously framed as being able to show not all Serbs had supported Radovan Karadzǐć(the RS's wartime president) and Slobodan Miloševic, or as being able to show what Serbs too had suffered during the war. These were indeed ways of 'recovering agency and selfhood' (Footitt, 2012: 227) during and after wartime, or of retrospectively introjecting a sense of agency into narrating one's experiences of them. In this set of narratives, however, they very rarely weighed more heavily than economic motivations for becoming interpreters. Moreover, the contradictory temporality of the peacebuilding missionsimultaneously drawn-out, stagnating, and always potentially about to end made it harder and harder to conceive of oneself contributing to reconstruction through facilitating international peacebuilding, showing how a lack of macropolitical agency overshadowed the micropolitical agency of language mediation itself.
Peacebuilding and agency in the post-Dayton 'waiting room' The social, political and economic structures within which locally-recruited workers are embedded, and how peacebuilders' attempts to construct a more stable post-conflict order have altered these (or not), are necessary but often disregarded context for discussing language intermediaries' agency in international intervention, even at the micropolitical level. In BiH, this context is inseparable from critiques of the Dayton system. Among the many criticisms of this settlement are that it institutionalized the ethnopolitical logic of the war itselfand many of the same elitesinto the post-conflict constitutional system without any mechanisms for Bosnians to contest it (Mujkic, 2007); it enshrined the gains of dubious clientelistic privatization of state socialist property (Kurtovic, 2015;Lai, 2016); and, by institutionalizing patriarchal ethnonationalism while abolishing state socialist quotas for women representatives in political assemblies, it failed to deliver even the liberal interpretation of gender empowerment that interveners professed (Björkdahl, 2012), far less any more transformative gender justice (O'Reilly, 2016). These dimensions of citizenship and participation should also be considered in a holistic view of locally-recruited language intermediaries' agency in post-conflict peacebuilding.
Though hardly any interpreters I interviewed in 2009-10 expressed satisfaction with the country's political or socio-economic direction, there was no strong evidence that interpreters as an occupational group were or desired to be agents of alternative politics, even though some individuals certainly interleaved their personal narratives with critical political commentary. One of these was a woman who introduced herself as the daughter of a mixed Muslim-Serb marriage, 9 and had had to move her family between three different central Bosnian towns (starting to work as an interpreter for British troops in the third town in 1997) before moving to Banja Luka for a better-paid administrative role at the divisional headquarters. Her own perception of Britain was as a country that handled national and religious differences better than Bosnia, in that she 'never thought, "Is Carl or Danny or Ben Protestant or Catholic or whatever?"' while working with British soldiers, whereas the equivalent information in BiH would, outside consciously 'post-ethnic' spaces, affect everyday social relations in BiH almost all the time. Working for British forces, in this context, had felt like 'some sort of […] personal sanctuary', where she could afford to set aside ethnicity as an identity category when dealing with others, and where she as a person who did not fit into any one ethnic category was not being ethnically profiled herself. 10 This was perhaps closer to the slight distance from ethnopolitics that wartime interpretersespecially young men who for reasons of ethnicity, politics and/or sexuality did not want to fightcould have achieved while working for UNPROFOR.
For this woman, speaking in 2010, the corruption, clientelism and ethnonational chauvinism she perceived in the leaders of the three main ethnonationalist parties were all aspects of the same malaise that was preventing Bosnians from living in a 'normal' country, and which was the root cause of the deficiencies she had encountered in the Bosnian health system (including lack of resources and widespread reliance on patients having to use informal practices to obtain treatment). Then able to hope that '[m]aybe with membership in the EU, things will change, because we'll be able to go abroad and see how people live there […] [and] see that what we have here is not good, is not healthy, it should be changed,' she anticipated that greater freedom to travel might permit more Bosnians to see what she had seen by working in spaces governed by the cultural and institutional norms of the liberalizing British military. 11 Yet the situation of 'mature' post-Dayton BiH is defined by such premises' perpetual deferment: BiH was no closer to EU membership by 2014, the year of the 'plenum' protests, 12 or indeed by the time of writing.
Social anthropologists have led the way in explaining how this situation affects everyday social relations in contemporary BiH. The condition of 'yearnings in the meantime' with which Jansen (2015) characterizes everyday social relations in contemporary BiH and individuals' understanding of what futures might be imaginable for themselvesthe contexts in which they elaborated their narratives of selfhoodoffers, after more than twenty years of 'peacetime', a broader structural perspective 9 'Muslim' in (post-)Yugoslav senses is an older ethnonym for 'Bosniak'. This paper uses Bosniak elsewhere, but uses 'Muslim' here to convey the speaker's self-positioning. 10 Interview, Banja Luka, 14 May 2010.

11
Ibid. 12 This public mobilisation occurred across cities in the constitutional entity known as the 'Federation of Bosnia-Herzegovina', two RS cities (Banja Luka and Prijedor), and the autonomous Brcǩo District. Previous manifestations demanding alternatives to Dayton had included the 2012 protests in Banja Luka against the private development of a city park, the 2013 'JBMG' protests in Sarajevo in 2013 (against the Dayton constitution's capacity to be exploited for causing ethnopolitical parliamentary deadlock, which was even preventing birth registrations) (Majstorovic, et al. 2015).
around the perpetual uncertainty that intervention agencies' locally-recruited workers faced. Today's Bosnian left commonly argues that the EU and other Euro-Atlantic institutions have left BiH in a perpetual semi-periphery of Europe by requiring it to work towards accession requirements while offering no transformative solutions, only further neoliberal measures such as the July 2014 'Compact for Growth and Jobs', offered as a response to public demands for radical political and economic reform (Majstorovic, et al. 2015). Viewed critically, the EU appears to keep BiH simultaneously in a 'swamp' of stagnation and a 'waiting room' of conditionality that is always just around the corner from being met, while spending the bare minimum on reform, and without unsettling the political and economic structures many Bosnians blame for the country's predicament (Jansen, et al. 2017: 23).
From this perspective, the precarity of local employment with foreign intervention agencies (which admittedly brought relative material advantages while workers were employed) did not even simply reflect the transnational casualization of labour, or international unwillingness to commit to the long-term presence of a mission that would eventually be wound down when BiH had stabilized. In 2009-10, the small number of EU and NATO troops then in BiH might still have appeared to be the last transitional phase of a dwindling mission; the more time passed, the more the remnants of the international peacebuilding structure seemed an expression of what post-conflict BiH would simply be, without any opportunity for citizens to participate in a democratic process that would change the conditions of the system in which they were living. This disempowering context and lack of democratic agency was the setting in which most past and present locally-recruited interpreters had been driven to become interpreters at all.

Conclusion
Since conducting the Languages at War interviews in 2009-10, two main developments with implications for conceptualizing language intermediaries' agency have been 'everyday' and micropolitical theorisations in peacebuilding, and the increasingly disempowering context of 'mature' post-conflict BiH, to which peacebuilding has contributed. Within the translation and interpreting process of peacebuilding, locally-recruited language intermediaries exercised significant, yet often unrecognized, agency. However, any 'local agency' they represented in broader peacebuilding processes was heavily circumscribed (though 'local' is too simple a category to convey the complex forms of disempowerment and social liminality experienced by interpreters and other locally-recruited employees). There was however some overlap between the stratum of (ex-)interpreters and certain 'local spaces of agency' (Kappler, 2014: 125) in BiH. The 'post-liberal peace' that optimistic critical scholars of Bosnian peacebuilding hoped could emerge from the interaction between extraparliamentary civic politics (mobilized primarily through arts and culture until 2014) and top-down mechanisms (Belloni, et al. 2016: 47) did involve a milieu of activists in which it was not uncommon for individuals to have worked as interpreters for a few years. This was a consequence of how their language-intermediary roles intertwined with their broader social positionality, since the social and intellectual backgrounds where young Bosnians were most likely to have acquired the competence in spoken foreign languages that interpreters needed (Kelly & Baker, 2013: 119) also loosely correlated with civic activism and 'post-ethnic' politics.
On the micropolitical and interpersonal level of peacebuilding, on the other hand, interpreters exercised an agency that peacebuilding research has largely failed to recognize. In encounters across the many social and ethnicised boundaries they crossed during their work (see Baker, 2015a), they engagedlike all Bosniansin the forms of everyday peacemaking activity one would expect to see in divided societies (Mac Ginty, 2014), and their competence in these everyday social practices had the potential to affect the outcome of an encounter for peacebuilders as well as everyday relations in interpreters' immediate and intimate lives. As active mediators of how peacebuilders gathered and produced knowledge, locally-recruited interpreters exercised agency on a deeply micropolitical scale. But the structural lack of agency arising from BiH's post-conflict peace settlement is likely to have overshadowed this everyday interpersonal agency more and more in how such intermediaries conceived of their role.

Funding
The interviews discussed in this paper were collected during the 'Languages at War: Policies and Practices of Language Contacts in Conflict' project, funded in 2008-11 by the Arts and Humanities Research Council.