Multilingualism in performance appraisal interviews: strategies to bridge language gaps

This paper explores three multilingual strategies to bridge the lack of a shared first language as they are used during performance appraisal interviews, namely English as a business lingua franca , receptive multilingualism, and a lay interpreter. The study is based on authentic video-recorded performance appraisal interviews at a globally active Belgian company, and follow-up interviews with the managers. The analysis focuses on how the company deals with multilingualism from a management perspective, how the managers perceive these multilingual strategies from a language ideological perspective, and how these language ideological beliefs shape the language practices of the performance appraisal interviews. We conclude that this study provides novel insights into how globalized companies deal with multilingualism in the high-stakes and potentially sensitive interactional context of a performance appraisal interview by underlining the intertwined nature of language ideological beliefs and language practices as part of a broader multilingual language management strategy


Introduction
As a result of late modern globalization, companies and their employees cross linguistic, professional, and national boundaries on a daily basis (Appadurai, 1996), and the ways in which companies and employees make use of language are therefore changing continuously.A crucial and recurrent interaction in the modern workplace is the performance appraisal interview (henceforth: PAI), a high-stakes encounter defined as "recurrent strategic interviews between a superior in an organization and an employee that focus on employee performance and development" (Asmuß, 2008, p. 409).Although a growing body of research has focused on this type of institutional interaction, most of the existing studies that are based on authentic empirical data are situated within specific nation-state frameworks where the interviews primarily take place in the respective dominant language (e.g., Asmuß, 2008;Mikkola and Lehtinen, 2014;Nyroos and Sandlund, 2014;Pälli and Lehtinen, 2014;Van De Mieroop and Schnurr, 2014;Van De Mieroop and Vrolix, 2014;Van De Mieroop and Schnurr, 2017;Meinecke and Kauffeld, 2018;Van De Mieroop and Carranza, 2018;Bowden and Sandlund, 2019;Lehtinen and Pälli, 2021).As such, the interactional specificities of multilingual PAIs remain relatively unknown to date, as well as how institutional interlocutors perceive them and how they fit into a broader language management strategy as part of an organization's general language policy.
De Malsche et al. 10.3389/fcomm.2024.1343267Frontiers in Communication 02 frontiersin.org Against this background, this paper examines the language practices, language management, and language ideological beliefs (Spolsky, 2004(Spolsky, , 2009) ) present during the PAIs at GlobalCorp (pseudonym), a small-sized service-oriented Belgian company that is active in 59 countries across the world and thus employs speakers with different first languages (L1s).More specifically, we examine videorecorded PAIs between managers who work at the company's headquarters in Belgium (L1, i.e., first language: Dutch) and sales agents who work for the company from all around the world (L1: not Dutch).During these interviews, the interlocutors make use of multiple multilingual strategies to bridge the language gap, namely English as a business lingua franca, receptive multilingualism, and lay interpreting.Additionally, the dataset includes two follow-up interviews with the managers responsible for the performance appraisal processes at GlobalCorp to better understand these multilingual processes from their insider perspectives.On the basis of this dataset, we adopt a qualitative interactional approach to examine (i) how the multilingual strategies used during PAIs at GlobalCorp fit into their general language management strategy, (ii) how the managers perceive the use of these strategies during PAIs, and (iii) how the managers' language ideological beliefs manifest themselves in the language practices of the PAIs themselves.
In the next section, we reflect on the role and influence of globalization on corporate language policies, particularly in the form of workplace multilingualism, to contextualize why we adopt a globalized and multilingual perspective on PAIs.We then elaborate on our methodology, research objectives, and qualitative analytical framework in Section 3. The analysis in Section 4 first provides insight into GlobalCorp's general management strategy and then focuses on the three multilingual strategies used during the PAIs specifically.Finally, we conclude that this study provides novel insights into how globalized companies deal with multilingualism in the high-stakes and potentially sensitive interactional context of PAIs by underlining the intertwined nature of language ideological beliefs and language practices as part of a broader multilingual language management strategy.
2 Language policy and multilingual interactions in a globalized workspace Following Spolsky's (2004Spolsky's ( , 2009Spolsky's ( , 2019) ) theoretical model of language policy, each (corporate) language policy is built on and comprises "three independent but interconnected components" (Spolsky, 2019, p. 326), i.e., language beliefs, language practice, and language management.The first component of language beliefs is described as "the values or statuses assigned to named languages, varieties, and features"; language practices are defined as "the observable behaviors and choices -what people actually do"; and the third aspect of language management comprises "the explicit and observable effort by someone or some group that has or claims authority over the participants in the domain to modify their practices or beliefs" (Spolsky, 2009, p. 4).In a globalized setting, most, if not all aspects of workplace communication have undergone complex changes over the past few decades (Argenti, 2006), and in light of such changes, most language policies have been argued to include a balancing act between finding an efficient way of communicating internally, which is often encouraged in the form of a common corporate language, and communicating externally with customers, which often requires more than one language (Angouri and Miglbauer, 2014).
In light of globalization, existing research on corporate language policies has focused strongly on the use of English as a lingua franca (ELF) or, more specifically, English as a business lingua franca (BELF, see Kankaanranta and Louhiala-Salminen, 2013;Kankaanranta and Louhiala-Salminen, 2018) for internal communication in the workplace.ELF, in a broad sense, is conceptualized as a multilingual practice or "multilingua franca, " rather than a variety of English (Jenkins, 2015), and can be considered both "the consequence and the principal language medium of globalizing processes" (Jenkins et al., 2011, p. 303).BELF is then defined as ELF as it is used by professionals in the domain of international business (Kankaanranta and Louhiala-Salminen, 2018), and has been specifically characterized by the resourcefulness and flexibility of its speakers to dynamically accommodate to different interlocutors in order to establish mutual understanding (Louhiala-Salminen et al., 2005;Rogerson-Revell, 2008;Köster, 2010;Kankaanranta and Louhiala-Salminen, 2013;Kankaanranta and Lu, 2013) and by an emphasis on intercultural comprehension, cooperation, accommodation and recipient design (Jenkins et al., 2011;Cogo, 2012;Mustajoki, 2017).However, research has also shown that the implementation of English as a common corporate language often does not solve the problems that linguistic diversity can pose in a multilingual workplace (Angouri, 2013;Angouri and Miglbauer, 2014;Lønsmann and Kraft, 2018;Sanden and Lønsmann, 2018), thereby underlining that a one-size-fits-all solution to corporate language policy does not exist (Sanden, 2016).Taking into account the balancing act between internal and external language policy needs, the use of different languages can be encouraged by companies if the use of these languages contributes to or results in financial advantages, particularly in the tertiary sector (Heller, 2010;Park and Wee, 2012;Angouri and Miglbauer, 2014); however, research has shown that managing this type of linguistic diversity can be challenging in practice (Fredriksson et al., 2006;Ehrenreich, 2010;Angouri and Miglbauer, 2014).
Next, to or in addition to BELF, a number of other strategies can be used to bridge the lack of a shared L1 in spoken communication (see Zeevaert, 2007;Lüdi, 2013 for an overview).One option is to make use of a lingua receptiva or receptive multilingualism, which entails that different interlocutors express themselves in their own preferred languages and at the same time have at least receptive proficiency in the language(s) that the other interlocutor is using (Zeevaert, 2007).Making use of these "asymmetrical competences" has been argued to benefit the efficiency as well as the fairness of the interaction (Lüdi, 2013).Another frequently applied strategy to bridge the lack of a shared language in all types of institutional encounters is the recruitment of an interpreter (Raymond, 2017).This can either be a professional interpreter, i.e., a trained, certified interpreter such as a sworn or community interpreter, or a lay interpreter, i.e., someone who is proficient in the relevant languages but has not been trained to act as an interpreter.These different strategies can be conceptualized separately, but in reality often overlap and intersect in multilingual interactional business settings.
The use of multilingual strategies forms part of an organization's broader language management strategies, and the thought processes behind when, how, and why these different strategies are used are indicative of the language ideological beliefs that the interlocutors adhere to.Language practices thus form a central locus of corporate language policies, and they become particularly relevant in interactions where the interlocutors do not share the same L1.Recurrent examples of such interactions include meetings, email communication, as well as more informal encounters such as coffee or lunch breaks.In this paper, we focus on the recurrent interaction of PAIs, as the evaluation of the performance and development of an employee makes this type of interaction a high-stakes and potentially sensitive encounter (Asmuß, 2008).This arguably results in an increased need for mutual intelligibility and successful information exchange, which can be complexified when the interlocutors do not share an L1.
Given their crucial role for both the employee and the employer, there is no lack of research on PAIs (see Asmuß, 2008 for a literature overview), and, especially over the past decade, scholars around Europe have examined these high-stakes interactions from discourse analytical perspectives on the basis of authentic empirical data, including studies from Sweden (Sandlund et al., 2011;Nyroos and Sandlund, 2014;Bowden and Sandlund, 2019), Germany (Meinecke and Kauffeld, 2018), Finland (Mikkola and Lehtinen, 2014;Pälli and Lehtinen, 2014;Sorsa et al., 2014;Lehtinen and Pälli, 2021), Denmark (Asmuß, 2008(Asmuß, , 2013;;Scheuer, 2014), Belgium (Van De Mieroop andSchnurr, 2017;Van De Mieroop and Carranza, 2018) and the Netherlands (Van De Mieroop and Schnurr, 2014;Van De Mieroop and Vrolix, 2014).The interactions in these studies primarily occur in the respective dominant language of specific nation-state frameworks and have significantly contributed to demystifying the interactional "black box" of PAIs (Clifton, 2012) through foci on facework, the (co-)construction of knowledge, leadership, and employeeship, and the interplay between talk and text during PAIs.However, despite the increasing globalized nature of corporate contexts, no studies to date have topicalized multilingual PAIs, i.e., PAIs in which interlocutors do not use their L1 to communicate.This study therefore aims to add such a globalized perspective to this existing body of empirical research on PAIs through a case study on the multilingual practices, beliefs, and management present during the PAIs at GlobalCorp as part of the company's broader language policy.

Methodology and research objectives
The research context for this study is a small-sized Belgian company pseudonymized as GlobalCorp.The company has approximately 25 employees working at their headquarters (henceforth: HQ) in Flanders, the Dutch-speaking part of Belgium, and operates in the tertiary sector in 59 different countries around the world.Instead of having a central point of communication at HQ for its clients, GlobalCorp employs 26 sales agents who are located around the world, each of whom represents a specific geographical market and is responsible for facilitating sales and maintaining contact with clients in their region.To date, GlobalCorp does not have an explicit language policy document which stipulates how to manage the multilingualism which results from this geographical distribution of employees.
For this study, we focus mainly on the communication between HQ and the sales agents at GlobalCorp, particularly during the performance appraisal process of the sales agents.The dataset analyzed for this study consists primarily of 7 video-recorded PAIs which were conducted via telecommunications software in 2021 and were part of the first round of structured PAIs ever conducted for sales agents at GlobalCorp.The total length of the dataset is 6 h and 47 min, with an average length of 58 min per video-recorded interview.Ethical permission to collect this dataset was provided by the Ethics Committee for the Social Sciences and Humanities of the University of Antwerp, and in line with this, all participants provided written informed consent prior to the recording of the interviews, and all data were pseudonymized during transcription. 1ELF was used as the primary multilingual strategy to bridge the lack of a shared first language in five of the interviews, one interview was conducted primarily through the use of receptive multilingualism, and one interview was mediated by a lay interpreter.It is relevant to note that although BELF was not considered the primary multilingual strategy for the latter two interviews, it was still present, as the managers spoke English in all of the interviews under study.Each interview took place with at least three interlocutors: the main interviewer (Manager 1, L1: Dutch) who is primarily responsible for asking most of the questions; a second interviewer (Manager 2, L1: Dutch) who is primarily responsible for setting up the report after the interview; and the interviewee (Agent, L1: not Dutch) whose performance is being evaluated.During the interview which was mediated by a lay interpreter, the interpreter functioned as a fourth interlocutor.
Manager 1 and Manager 2 are responsible for the performance appraisal processes at GlobalCorp and are the same in each PAI.After a preliminary analysis of the interactional data, the first author conducted and audio-recorded two separate follow-up interviews with them to gage their emic experiences with and perceptions of the performance appraisals at GlobalCorp, including language and multilingualism as part of these processes.The interview guide for these semi-structured interviews can be found in Appendix 1.
Building on Spolsky's theoretical model of language policy which comprises "three independent but interconnected components" (Spolsky, 2019, p. 326), we formulate three research objectives: 1. We aim to examine how the multilingual mediating strategies used during the PAIs at GlobalCorp, i.e., BELF, receptive multilingualism, and lay interpreting, fit into the company's general language management strategy.2. We aim to shed light on the ways in which the managers talk about and perceive the use of BELF, receptive multilingualism, and lay interpreting during PAIs as reported on during the follow-up interviews, as their reflections provide insight into the language ideological beliefs they adhere to with regards to the use of these different multilingual strategies. 2 3. We aim to examine these language ideological beliefs in light of the interactional language practices of the PAIs, and we focus in particular on (potential) miscommunication during the relevant to reach a full understanding of language policy and practices at GlobalCorp, our focus is limited to the managers' perceptions and beliefs because the data access agreement with GlobalCorp precluded us from being in direct contact with the agents.
De Malsche et al. 10.3389/fcomm.2024.1343267Frontiers in Communication 04 frontiersin.org PAIs, as we argue that the managers' risk assessment of a particular strategy shapes the way in which they prevent, signal, and/or repair (potential) miscommunication in interaction, thereby reflecting the language ideological beliefs they adhere to with regards to that strategy.
To achieve these research objectives, the first author first conducted a discursive in-depth turn-by-turn analysis (Van De Mieroop and Vrolix, 2014) of each PAI with a focus on instances of (potential) miscommunication using the coding software NVivo.To do so, she identified prevention and signaling strategies on the basis of previous research on miscommunication in multilingual spoken interaction (Linell, 1995;Vasseur et al., 1996;Mauranen, 2006), as well as examined the ways in which the interlocutors repair (potential) miscommunication (Schegloff et al., 1977).Miscommunication is generally considered a "slippery concept" (Coupland et al., 1991, p. 11) which is closely related to concepts such as "misunderstanding, " "trouble talk, " "conflict talk, " or "communicative breakdown" (Linell, 1995).For the purposes of this study, miscommunication is defined as "talk non-deliberately generating or mobilizing and sometimes leaving discrepancies between parties in the interpretation or understanding of what is said or done in the dialog" (Linell, 1995, p. 176-177).
Although many examples of miscommunication in the dataset were ambiguous in terms of why they occurred, the examples discussed in the analysis are chosen on the basis of their clear link to the language gap between the interlocutors, thereby aiming to exclude other sources of potential miscommunication, such as internet connection.The excerpts in the analysis are transcribed according to a simplified version of the Jeffersonian transcription method (Jefferson, 2004) and the transcription conventions can be found in Appendix 2.
To link the findings on language practices with the managers' language ideological beliefs and language management strategies, we then adopted an interactional sociolinguistic approach to foreground "what people do versus what they say they do" (Marra et al., 2022), as part of which we analyzed and coded the follow-up interviews with the managers on the basis of a topic-based categorization in the second phase of the analysis.All sequences relating to language, multilingualism, language management, language policy, the use of English/BELF, receptive multilingualism, or lay interpreting were considered relevant to and included in this analysis.

Language practices, management, and beliefs in multilingual performance appraisal interviews at GlobalCorp
In the following section, we focus on the first research objective relating to language management as we provide an overview of GlobalCorp's general language management strategy and discuss how the multilingual PAIs are part of it.In the subsequent sections, we then focus on the second and third research objectives relating to language ideological beliefs and language practices by examining the specific language ideological beliefs that the managers adhere to related to the use of BELF, receptive multilingualism, and lay interpreting, respectively, during the PAIs, and how these beliefs manifest themselves in the language practices of the multilingual interactions.

GlobalCorp's language management approach
In terms of formality, GlobalCorp's approach to language management and multilingualism has not been codified explicitly in any language policy document, thus resulting in a generally implicit language policy (Shohamy, 2006).Similar to other globally active companies, it became clear during the follow-up interviews with management that this implicit language management strategy involves a balancing act between their internal language policy, which strives for efficiency in the form of a common corporate language, and their external language policy, which aims to reach as many potential clients as possible in the language of their preference (Angouri and Miglbauer, 2014).
Instead of working with a central point of communication at HQ in Belgium, the managers at GlobalCorp explained during their follow-up interviews that they believe it is the agents' knowledge of the local market, local culture, and local language(s) in their targeted markets around the world that makes the sales agents more approachable for potential clients as company representatives, thereby resulting in more clients and thus more sales.In other words, the agents' proficiency in local languages and knowledge of the local culture is considered marketable and therefore encouraged for external communication as part of a "transnational business model" which ties together "local responsiveness" with "strong global direction" (Feely and Harzing, 2003, p. 37).Manager 2 explains that "from a commercial viewpoint, [we] always adapt to the language of the client whenever possible"3 , and GlobalCorp's general external language policy can thus arguably be considered flexible and client-oriented.
The company's approach to external communication differs from their internal approach, as the managers explain that Dutch is the primary language used for all communication between the employees at HQ (including management), while English forms the baseline of the internal communication between HQ and the agents abroad.Multilingual strategies other than BELF are construed as "pragmatic solutions"4 which are only called upon when BELF does not suffice to achieve mutual understanding.In terms of their own linguistic repertoires, both managers have Dutch as L1 and both are proficient in English, with Manager 1 reporting intermediate to high proficiency and Manager 2 reporting near-native proficiency.Additionally, they both report intermediate receptive proficiency in German and Manager 2 also has intermediate proficiency in French.Although both managers report on using their multilingual repertoires in different ways and to different extents in the workplace, Manager 1 explains that from a language management perspective, "I actually hope that in time, we can do it with everyone in English of course, but yeah, unfortunately not yet at the moment." 5 As such, when it comes to internal communication between HQ and the agents abroad, GlobalCorp currently adopts a flexible approach to language management, but in line with other European businesses operating on a global scale (Gunnarsson, 2014), they hope to eventually adopt English as a common corporate language with all the sales agents, sharing an implicit belief that this is the best solution to solving any problems related to workplace multilingualism.

English as a business lingua franca in PAIs
Despite their overall positive attitude toward the use of English as a common corporate language, both managers adopt a more nuanced view with regard to the benefits and disadvantages of using English as a primary multilingual strategy to bridge the lack of a shared L1 specifically during the PAIs.During the follow-up interviews, Manager 1 acknowledges that it is more difficult for agents to express themselves in a language that is not their own L1, and Manager 2 shares that he believes it would be best if everyone could use their own L1, especially in high-stakes and potentially sensitive encounters such as PAIs.However, both managers also seem to consider English to be a marker of "intergroup neutrality, " meaning that "no member of the group is particularly privileged" when everyone is required to use a language other than their L1 (Park and Wee, 2012, p. 146).The interpretation given to this 'neutrality' differs however slightly between both managers, as for Manager 1, it concerns all interlocutors including himself, since he is also not an L1 speaker of English and thus explains that he shares the disadvantage of not using his own L1 with the agents.Manager 2 arguably adopts a more top-down perspective when he explains that they prefer to opt for English "… because then we can put everyone on equal footing" 6 , and in doing so, he reflects mostly on how the managers aim to ensure equal treatment for all the sales agents as part of this evaluative process, rather than including the managers themselves in his conceptualization of 'everyone' .
Overall, Manager 2 also adds that he considers the use of English to be commonsensical because "English is the only lingua franca that, I mean, it is the only one that you can actually assume or reasonably 6 Dutch original: "…omdat we dan iedereen op een gelijk niveau zetten."expect someone else, other people to also know and understand." 7 In sum, the managers seem to share the widespread language ideological belief regarding the hegemonic status of BELF in international business (Gunnarsson, 2014) and believe that it offers a 'neutral' solution to bridge the language gap between the managers and sales agents at GlobalCorp because the disadvantage of not having English as their own L1 is shared by the agents.Because of this assumed neutrality, English forms the baseline of GlobalCorp's internal language management strategy for all communication with the sales agents, and is therefore also used as the primary multilingual strategy in the majority of the PAIs.
During the PAIs, we find that the managers' awareness of the shared disadvantage of not being able to speak their own L1 also shapes their language practices.An example of this occurs with a sales agent responsible for a Western European market (L1: French) when they are discussing his home office space.Prior to the PAI, the agent did not have a proper home office space and his laptop had recently broken as well, both of which can be considered detrimental to his performance as a sales agent.In Excerpt 1, he announces that he is working on solving these problems.
In turn 3, the agent experiences word-finding difficulties regarding the space where he plans on having his home office, saying that they "put some room" somewhere.In turn 4, Manager 2 offers what Mauranen (2006), p. 137 calls an interactive or proactive repair to prevent miscommunication and co-construct the agent's explanation to find out where exactly he is making room.This proactive repair is unsuccessful, as the agent replies in turn 5 that he did not mean the garage, but another "chamber." Manager 2 seems to rely on his own proficiency in French to interpret this use of the word "chamber" as a false friend from French, and subsequently translates it to "a room in the house" in turn 6, thereby offering another interactive repair.It is relevant to note that although the agent had already used the term 'room' in turn 3, he did so in a different sense for which two separate lexical items are used in French; in turn 3 it would be translated to French as 'place', whereas in turn 5 it would be translated as 'chambre'.As such, the agent knows the English word 'room' , but he might 7 Dutch original: "Engels is de enige lingua franca die, allez, dat is de enige dat eigenlijk dat je vanuit kunt gaan of van kunt verwachten redelijkerwijs dat een ander, dat andere mensen dat ook gaan kunnen begrijpen."be confused by or unaware of its double meaning in English.The agent does not react to Manager 2's interactive repair in turn 7 but continues talking about the office, thereby marking an implicit acceptance of the repair offered in turn 6.In this first part of the sequence, Manager 2 thus accommodates to the (lack of) English proficiency of the agent by trying to help him find a solution to the word-finding difficulties he is experiencing in the form of two interactive repairs, thereby successfully solving the potential miscommunication.
A few minutes later, Manager 1 is providing feedback for the agent when he adds that he thinks this new home office will also help improve the agent's performance.
In doing so, Manager 1 repeats the word "chamber" instead of "room" in turn 38, arguably to accommodate the agent by using the same word that the agent had used earlier.It is possible that because Manager 1 does not have any proficiency in French himself, he interpreted the miscommunication in the first part of the sequence as the agent not knowing the word 'room' , instead of as word-finding difficulties related to confusion or unawareness regarding the double meaning of the word in relation to French.The manager ends this turn by asking a direct question that deictically links back to the term "chamber" ("any idea when it's when it's done?").This then results in miscommunication as the agent pauses and explicitly asks for clarification in turn 39, launching a repair sequence of which the word "chamber" is likely the trouble source.Manager 1 tries to repair the miscommunication in turn 40 by first repeating the false friend again and then offering the English alternative "office." Finally, in turn 41, the agent laughs and uses the term "room" in the sense of its French equivalent 'chambre', and the miscommunication is definitively repaired.This second part of the sequence shows that paradoxically, Manager 1's attempt to accommodate to the agent's linguistic proficiency in order to prevent miscommunication backfires and results in a repair sequence, the exact thing he was presumably trying to avoid by saying 'chamber' instead of 'room' .Moreover, as part of the agent's reply in turn 41, a second potential French-English false friend occurs, as the French term 'bureau' can translate to both 'desk' and 'office' in English.In saying that he will go to the store to buy "a big office, " the agent thus seems to refer to the former meaning of the term 'bureau', rather than the latter.Despite his lack of proficiency in French, it is likely that the manager was able to deduce on the basis of contextual information that the agent was referring to a desk, and in contrast to the previous false friend, this does not result in any discernible miscommunication, as the manager seems to use the accommodation strategy of 'letting it pass' (Firth, 1996) in turn 42.
In sum, despite the fact that the language gap is arguably rendered less visible through the use of English as a lingua franca, this sequence highlights both managers' awareness of the potential risks that might arise from not being able to speak one's own L1 in a high-stakes interaction such as a PAI, especially when it concerns the discussion of performance improvements.In the first part, when the agent wants to clarify that he aims to improve his performance by improving his home office space, Manager 2 makes an effort to accommodate to the agent's proficiency in English by offering proactive repairs on the basis of his own proficiency in French.In the second part, Manager 1 then provides positive feedback on the agent's effort to improve his performance, and thereby also makes an effort to accommodate by repeating the false friend that the agent himself had used only minutes earlier, and by letting another false friend pass.These examples are illustrative of the many and different ways in which the managers aim to establish mutual understanding during moments of feedback in the 5 PAIs where BELF was used as the primary multilingual strategy to bridge the lack of a shared L1.
This analysis and the interlocutors' linguistic behavior in the PAIs aligns with findings from previous research on (B)ELF which has argued that interlocutors are resourceful, accommodating, and cooperative in achieving mutual intelligibility (Louhiala-Salminen et al., 2005;Rogerson-Revell, 2008;Cogo, 2009Cogo, , 2012;;Firth, 2009;Jenkins et al., 2011;Mustajoki, 2017) and "exhibit a high degree of interactional and pragmatic competence" when dealing with (potential) miscommunication (Pitzl, 2005, p. 69).Although the managers share a language ideological belief in the hegemonic status of English as a neutral and fair solution to the lack of a shared L1 between the managers and the agents, the discourse strategies they use in interaction to prevent, signal, and repair (potential) miscommunication and achieve mutual intelligibility during moments of feedback also suggest awareness of the risks they associate with not being able to speak one's own L1, particularly in potentially sensitive interactions such as the PAI.

Receptive multilingualism in PAIs
One PAI with an agent responsible for a Western European market (L1: German) takes place primarily through the use of receptive multilingualism.During the follow-up interviews, Manager 1 explains that this agent had made it clear that he prefers to speak German whenever possible, and the managers are able and willing to accommodate to the agent's linguistic preference in the PAI context because of their own receptive skills in German.As a result, the managers speak English and the agent replies in German, with both parties occasionally and briefly code-switching to the other language Excerpt 1 -part 2 (00:15:13-00:15:30 out of 00:48:22) -interlocutors present: Manager 1, Manager 2, agent for a Western European market 38 Manager 1: …and I think a new chamber will be uh interesting for you so uh (.) any idea when it's when it's done?39 Agent: ( as an accommodation strategy (Cogo, 2009).Because of the managers' spoken use of English and the agent's receptive skills in English, (B) ELF is also present as a "multilingua franca" in this PAI (Jenkins, 2015), but it is not considered the primary multilingual strategy.Both managers express that they find the multilingual strategy of receptive multilingualism a bit unusual or what Manager 1 calls "unique, " but they believe it works well given the circumstances, particularly because it allows the agent to express himself in his own L1 in a potentially sensitive interactional context, tying in with the managers' beliefs regarding the use of one's own L1 as discussed in the previous section.Reflecting on their own proficiency in German, Manager 2 says: "I think that both (MANAGER 1) and I understand German well enough that misunderstandings would not occur, I think, "8 a statement which he hedges through the double repetition of "I think, " arguably reflecting some hesitation or doubt.Despite this implicit association with a potential risk of miscommunication and despite their own lack of productive proficiency in German, the managers thus prioritize the agent's comfort in this potentially sensitive and high-stakes workplace interaction.
During the PAI, the interlocutors' awareness of the unusual nature of this multilingual strategy seems to result in the frequent use of a number of discourse strategies to prevent (potential) miscommunication and accommodate to the use of different languages.For example, Excerpt 2 showcases Manager 1's alertness to the potential risks of receptive multilingualism when discussing a potentially sensitive topic.Leading up to this excerpt, they had been discussing the opportunity for the agent to visit a specific client in person, something Manager 1 considers important for the agent as part of maintaining good client relations.However, in turn 1, the agent explains that he actually does not think this is a good idea, as the client in question has not seemed interested in maintaining a good relationship with him in the past.
In turn 2, Manager 1 then initiates a repair sequence through the use of a specific metalinguistic question (Vasseur et al., 1996, p. 88) by stating that he has not understood the agent's prior utterance due to the use of a German expression in turn 1 ("auf etwas Wert legen").He asks Manager 2 if he knows what it means, thereby seemingly asking for a translation to Dutch rather than a repair from the agent.However, before Manager 2 can reply, the agent himself rephrases the expression in German in turn 3, albeit in a simplified way, as he leaves out the client's lack of willingness to be in contact with the agent and replaces it with a more neutral description of the client's lack of action.Manager 1's minimal reply in turn 4 indicates that he has understood and that the miscommunication has thus been successfully repaired, allowing them to move on.This type of specific metalinguistic question arguably "always results in better understanding, because it clearly, precisely and cooperatively triggers working sequences" (Vasseur et al., 1996, p. 88), and as such reflects Manager 1's alertness and clear orientation to finding an efficient solution to potential miscommunication at a sensitive moment in this multilingual interaction, namely when he suggests for the agent to do something, and the agent disagrees with him.
Similarly, Excerpts 3 and 4 in turn highlight the agent's awareness of the managers' relatively low proficiency in German as he proactively rephrases or translates words which the managers might not be able to understand while explaining something to them.Excerpt 3 is an example of the agent clarifying the term "günstig" to explain that this means the client bought the product for a cheaper price, using the sales term "Normalpreis" in his reformulation.Although we cannot know for certain that this clarification is tied to the agent's perception of Manager 1's proficiency in German, it is possible that this is the case.Excerpt 4 is more clearly tied to the agent's assessment of the managers' linguistic proficiency, as the agent briefly code-switches to get his point across by translating the German term "fertig" in turn 1 to the English term "finished" in turn 3.In both excerpts, these reformulations are unsolicited and arguably highlight the agent's awareness of the potential risks associated with the managers' linguistic proficiency (or lack thereof), which in turn seems to manifest itself in different discourse strategies to prevent miscommunication during the PAI.
The excerpts show that the interlocutors make use of a number of discourse strategies to ensure that miscommunication is prevented or that it is signaled and repaired efficiently when it does occur, including (but not limited to) during potentially sensitive moments such as disagreement.With this type of multilingual strategy, the language gap is rendered highly visible because the interlocutors are not only confronted with an interlocutor speaking in a language which they do not fully master, but they are also consistently replying in a language that is different from the language used in the previous turn, and this heightened risk awareness arguably shapes the language practices of both the managers as well as the agent during the high-stakes workplace interaction of the PAI.hiring a professional interpreter, partly because of its costliness, but also out of fear that sensitive corporate intel might be leaked as a result of working with a third party.This risk is then arguably mitigated by relying on a (relative) insider, such as a family member, a friend, or another colleague.In the case of the agent's spouse, the managers clearly trust her, both in terms of communicative interpreting skills and in terms of confidentiality, as she is not asked to sign any legally binding documents pertaining her role as an interpreter in the PAI.
Although both managers thus showcase a certain awareness regarding some of the potential risks involved with (lay) interpreting in general, neither of them indicates that something might have gone wrong during this PAI specifically, and they both defend the company's decision to use the spouse as an interpreter in a number of ways.Additionally, by sharing their beliefs that "nothing was lost" and by referring to the position of the interpreter as a "gap" that needs to be filled, they seem to implicitly adhere to the language ideological belief that during the PAIs at GlobalCorp, the interpreter functions as "a mere medium of transmission" (Knapp-Potthoff and Knapp, 1986, p. 153) who renders an original utterance from language A to language B as (nearly) equivalent.This common assumption is captured in the "conduit metaphor" (Reddy, 1979) and reflects a language ideological belief of "referential transparency" (Haviland, 2003, p. 764) which considers interpreters to be "invisible" or "machine-like conduits" (Berk-Seligson, 1990, p. 54;Pöchhacker, 2004; see also Angermeyer, 2015).
Despite this lack of perceived risk, the interlocutors do make use of a number of discourse strategies to signal and repair (potential) miscommunication so as to achieve mutual intelligibility, examples of which occur in Excerpts 5 and 6.
Both excerpts are examples of interactive repairs (Mauranen, 2006, p. 137) at moments when the interpreter is conveying positive aspects of the agent's performance and where the managers make an effort to co-construct a turn when the interpreter is experiencing word-finding difficulties in English.Manager 1 tentatively reformulates the interpreter's hand gestures with rising intonation in turn 2 of Excerpt 5 into a successful repair, which the agent is able to confirm himself despite his lack of proficiency in English due to the likeness of the English "long term" and the French "long terme." In Excerpt 6, Manager 2 makes use of his own proficiency in French to achieve mutual intelligibility for Manager 1 in the form of a translation to English in turn 2. Such examples underline the cooperative nature of achieving mutual understanding multilingually, particularly during the explicit discussion of the agent's performance.However, the 'double discourse' of original turns and subsequent renditions in interpreter-mediated interaction also allows researchers to observe "latent miscommunication" or "pseudo-agreements, " i.e., instances of miscommunication of which there are no traces in the interactional data and where the interlocutors believe that they have understood each other when, in fact, they have not (Linell, 1995).
Such instances of pseudo-agreements, albeit minimal and not extensive, occur multiple times in the interpreter-mediated PAI, for example when Manager 1 attempts to provide the agent with positive feedback on his everyday communication with HQ in Excerpt 7.
In turns 1 and 3, Manager 1 gives the agent exclusively positive feedback, emphasized by the word "good" which is repeated four times throughout his explanation, and it is formulated without any recommendations or advice on what he should or could improve on in the future.In doing so, he first refers to the agent in the third person, thus addressing the message to the interpreter, but then Excerpt 5 (00:07:20-00:07:27 out of 00:51:56) -interlocutors present: switches to addressing the agent directly in the middle of turn 3 by using the second person pronoun "you." In turn 4, the interpreter starts with providing a reduced rendition of this positive feedback ("keep doing what you are doing right now").She then continues with an expanded rendition in turns 4, 6, and 8, as she emphasizes the importance of replying to clients quickly and how he should approach doing so, certain aspects of which were not mentioned in the original utterances in turns 1 and 3. Additionally, the rendition of the feedback in turns 4, 6, and 8 is phrased by the interpreter as recommendations on how the agent could improve his communication in the form of a number of imperatives ("do so right away, " "tell them, " "you have to, " "stay true to your word"), this does not in any way reflect the positive emphasis of the feedback given in Manager 1's original utterances.In her renditions, the interpreter thus changes both the content and the connotation of Manager 1's original turns from praise to implicit criticisms.In turns 5, 7, and 9, the agent then signals agreement with this rendition, despite it not being an accurate reflection of Manager 1's original feedback, indicating pseudo-agreement between the primary interlocutors: the managers believe that the agent is agreeing with the positive feedback and praise, while the agent has actually agreed with the feedback, as well as advice and recommendations on how he could or should maintain his performance in the future.It is also relevant to note that Manager 2 does not interfere to repair this pseudo-agreement despite his own proficiency in French, though this could be due to part of the rendition being in Arabic, which he has no proficiency in.
Although there are also sequences in the interview where such mistranslations are intercepted and repaired by one of the other interlocutors, sequences such as Excerpt 7 highlight that miscommunication does occur and, when it does, remains invisible and therefore largely unrepaired during this PAI, including during crucial moments such as the communication of positive feedback.
In addition to mistranslation, certain sequences were not rendered at all, as is the case at the end of the PAI in Excerpt 8.
In turn 1, Manager 1 provides important detailed information regarding the next steps of the appraisal process, including how the report will be set up, that the agent is expected to read and sign the report, the timing of the next performance appraisal, and the timing of their next individual meetings.In turn 2, the interpreter provides only a rendition of the final question of Manager 1's turn without rendering any of the prior information.All of the information regarding the follow-up of the PAI thus remains uncommunicated to the agent, and a repair to this miscommunication is not initiated by either of the managers, although Manager 2 could understand the interpreter's (lack of) translation to French in turn 2, and both managers could arguably deduce on the basis of the length of turn 2 that the original turn had not been rendered in full.
Interpreter-mediated communication has been argued to be inherently more prone to miscommunication than unmediated interactions because mutual intelligibility between the primary interlocutors depends on the mediation and understanding of a third party (Wadensjö, 1998), and problems occur more frequently when the interpreter is not a trained professional (Angermeyer, 2015).This is highlighted in Excerpts 5 and 6 as the lay interpreter frequently experiences word-finding difficulties in English and requires help from the other interlocutors to complete her renditions.Yet despite the numerous risks associated with non-professional interpreters in the workplace (Raymond, 2017), they are still frequently used due to their low cost and relative ease of availability.
Although both managers at GlobalCorp express awareness on a meta-level over some of the general potential risks involved with using an interpreter, they do not reflect on any problems that might occur or have occurred in the past with this specific lay interpreter.When asked during the follow-up interviews, Manager 1 explicitly indicates that he believes "nothing was lost, " and Manager 2 reacts surprised when the first author eventually tells him that we observed mistranslations as part of our interactional analysis.This reveals that they were, in fact, unaware of the miscommunication that occurred during this PAI, even when they could have been alerted to it, for example by using Manager 2's own proficiency in French to intercept the pseudo-agreement in Excerpt 7 or by being alert to the length of the rendition of the interpreter in Excerpt 8.Such examples showcase Excerpt 8 (00:47:59-00:48:46 out of 00:51:46) -interlocutors present: Manager 1, Manager 2, agent for North African and Middle Eastern markets, interpreter 1 Manager 1: so good and and this this was uh the performance review so (.) now (MANAGER 2) will make a summary and uh we'll finish it and I will send it to you so you can sign also the document you can see what we wrote (.) and uh this we will do once a year uhh but next to that we will keep our meetings one on one of course uh once in six seven weeks or so uhh when it's needed of course in the summer it's not highly needed uh but like in august there's the start of ( that the managers seem to consider the prevention, signaling or repair of any potential miscommunication and the general achievement of mutual intelligibility to be primarily the interpreter's responsibility, rather than that of the primary interlocutors, and they only jump in to help when the interpreter indicates that she needs it, as is the case in Excerpts 5 and 6.This lack of perceived risk potential is arguably embedded in the managers' implicit language ideological belief that an interpreter provides (nearly) equivalent renditions of original utterances (Reddy, 1979), and as such, they assume an "ideal interpretation" in which the interpreter solves any potential miscommunication that might occur, which stands in contrast with the reality of the "actual performance" of the lay interpreter (Wadensjö, 1998, p. 103).

Concluding remarks
In light of the increasingly globalized and multilingual nature of corporate contexts worldwide and the institutional relevance of PAIs for workplace development, this paper has aimed to contribute a multilingual perspective to the existing body of research on PAIs by examining the language policy of a small-sized yet globally active Belgian company through a qualitative analysis of the multilingual strategies they use to bridge the lack of a shared L1 during PAIs.Building on Spolsky's three-pronged theoretical model of language policy (Spolsky, 2019), we have explored how these multilingual strategies fit into the company's general language management strategy, how the managers perceive the use of these different multilingual strategies during PAIs, and how their language ideological beliefs manifest themselves in the language practices of the PAIs themselves, particularly during sequences of (potential) miscommunication.
In terms of their general language management strategy, the follow-up interviews revealed that GlobalCorp currently adopts a multilingual "pragmatic and flexible approach to language use" (Louhiala-Salminen et al., 2005, p. 418) for their internal communication.However, similar to other international businesses in Europe (Gunnarsson, 2014), they aim to implement English as a common corporate language for all communication with their sales agents in the future, which we argued is rooted in their language ideological belief in the hegemonic status of English in international business as a neutral and fair solution to the lack of a shared L1.
For the PAIs, we found that they currently make use of three multilingual strategies to bridge the lack of a shared L1 between the interlocutors, namely BELF, receptive multilingualism, and a lay interpreter.We found that the decision to use any of these strategies is rooted in specific practical considerations as well as language ideological beliefs, which result in differing language practices during the PAIs.Despite the managers' general language ideological belief in the problem-solving potential of English, the language practices in the PAIs where BELF is used as the primary multilingual strategy showcased some risk awareness regarding communicating in a language other than one's own L1 during a high-stakes and potentially sensitive workplace encounter.The managers expressed more risk awareness for the use of receptive multilingualism, which is reflected in the language practices of all interlocutors in the PAI through a number of prevention and repair strategies, including during potentially sensitive moments such as disagreement.Finally, the findings on the lay interpreter-mediated PAI seemed to lay bare a discrepancy between the managers' perceived risk of this multilingual strategy and the actual language practices of the PAI, which was tied to the managers' implicit language ideological belief that an interpreter provides equivalent renditions of original utterances, and we argued that such a discrepancy can at times result in unrepaired miscommunication during crucial feedback and information moments of the PAI, as showcased in the analyzed excerpts.Although these three strategies were analyzed separately, BELF was present throughout all the PAIs, thereby underlining its function as a "multilingua franca" that is always "in the mix, " even when it is not the primary strategy being used (Jenkins, 2015, p. 74).Moreover, the intertwined nature of these different strategies then highlights that the multilingualism of these interactions cannot be neatly categorized as different multilingual strategies, but rather that these strategies overlap and intersect in authentic multilingual interactions.
In sum, we have offered novel insights into how a small-sized yet globally active company deals with multilingualism on the levels of language management, language beliefs, and language practices as part of their performance appraisal process, thereby contributing an explicitly globalized perspective to the growing body of research on PAIs.In doing so, we highlighted that in the interactional context of multilingual PAIs, the language beliefs regarding the potential risks of specific multilingual strategies can shape the interlocutors' language practices in interaction.For strategies which are associated with a risk of miscommunication, such as BELF and receptive multilingualism, we found that the interlocutors make an effort to prevent, signal, and/ or repair (potential) miscommunication, particularly during moments in the PAI which could be sensitive in nature, such as during the discussion of performance improvement and the communication of feedback.In light of these findings, we would argue that it might be because of this high-stakes and potentially sensitive nature of the PAI setting that the necessity of achieving mutual intelligibility is heightened, thereby contributing to these interactional efforts.However, a lack of perceived risk and communicative responsibility can similarly shape language practices in the form of a lack of interactional effort to prevent miscommunication, as seen in the lay-interpreted PAI, resulting at times in miscommunication at crucial and sensitive moments.Methodologically, we argue that the qualitative triangulation of authentic empirical data and emic interview data has been crucial to achieving these insights, and that similar approaches can further contribute to achieving multifaceted and detailed insights on the role and function of language and multilingualism in workplace interactions and corporate settings as a whole.
Reflecting further on the importance of the managers' emic perspectives, a fruitful area for future research could be to investigate their language ideological beliefs from a more critical discourse analytical perspective, for instance by investigating further what it means for managers to achieve 'linguistic equality' (Tonkin, 2015) or fairness in a multilingual workplace setting where everyone is disadvantaged if they cannot speak their own L1, yet not equally so, as the degree of disadvantage is dependent on an individual's personal linguistic repertoire.By further examining these language ideological beliefs and relating them back to a multilingual company's language management strategies and language practices in high-stakes interactions such as the PAI, we can gain a better understanding of what linguistic equality and fairness mean in globalized corporate contexts.