Participant role shifting in minority language institutional interactions: COVID-19 contact tracing telephone calls as complex spaces of linguistic non-understanding

1 Researching Flemish COVID-19 contact tracing in minority languages as spaces of linguistic non-understanding

One of the main concerns and goals of public authorities during the outbreak of the COVID-19 pandemic was to inform and advise citizens on health-related decision-making worldwide. As a result, COVID-19 brought about an almost unprecedented flow of health-related communication, which arguably contributed substantially to the public health awareness and health literacy of many people around the globe. This evolution can naturally be attributed to the instrumental importance of centralized and decentralized public health initiatives, both to build knowledge about infection rates, and to inform and sensibilize citizens of the dangers of COVID-19 and the corresponding safety measures in place at a given time. One example of such an initiative is government-mandated telephone contact tracing, which, in Belgium, emerged at the time as a relatively novel, but strikingly upscaled institutional activity type (Levinson 1979) to combat the spread of the COVID-19 virus. Contact tracing itself of course was not new, with earlier uses in the contexts of Ebola outbreaks, the HIV/AIDS epidemic, and tuberculosis (Bafort et al. 2023), but the sheer scale at which large number of patients were contacted during COVID-19 was unprecedented. The shift to telephone contact tracing through call centers equally marks the upscaled nature of public health intervention. We will therefore focus on this specific type of contact tracing in this article, viz. telephone contact tracing conducted by call center operatives. A COVID-19 contact tracing interaction takes place when an infected person tests positive for a contagious virus (the so-called index patient, hereafter IP). This person is contacted by a contact tracer (CT), asked about their contacts and, in the COVID-19 context which we studied, the IP is provided with instructions and other information regarding safety and prevention measures (De Timmerman et al. 2023; Slembrouck et al. 2023).

The present article reports on data and analyses from an interdisciplinary research project on government mandated contact tracing telephone interactions in Flanders during the COVID-19 pandemic. From late 2020 until early 2022, we recorded and analyzed over 200 contact tracing calls between CTs and IPs. The resulting corpus consequently spans the second half of the COVID-19 pandemic, which was characterized by varying degrees of government-mandated prevention measures, ranging from complete lockdown to the advised wearing of mouth masks (for a detailed timeline, see Coppens et al. 2021). The main aims of the project were to map the interactional dynamics of these telephone conversations, and to help improve the institutional practice at hand. The project was conducted in close collaboration with the Flemish Agency of Health and Care, which oversaw the contact tracing operation in Flanders, and one of the call center firms which were responsible for the ‘on the ground’ implementation of the agency’s policy. In this way, we had direct access to both higher- and lower-level agents who played crucial roles in combatting the spread of COVID-19 in Flanders.

One of the interactional features of the institutional setting under scrutiny in this paper is the inclusion of and accommodation for various minority languages. Belgium’s federal language policy dictates that official communication between citizens and government institutions take place in one of the country’s three national languages: Dutch, French or German, depending on the region where the practice is situated. COVID-19 telephone contact tracing was institutionally organized country-wide and per region. Our focus lies on Flanders, i.e. the officially Dutch-speaking region. In practice, however, these and other, less formal, institutional interactions often accommodated for linguistic minority groups either through the use of a lingua franca, or through various forms of remote and on-site interpreting. This is particularly prevalent in various contexts of public service (cf. Roels et al. 2015), such as civil registry interactions (Vandenbroucke and Defrancq 2021), and medical consultations (Cox and Maryns 2021). Strikingly, during the pandemic, an exemption to the federal language laws was implemented for local governments in Flanders, allowing them at that time to spread information in languages other than Dutch (Coronavirus: Meertalige informatie – Agentschap Integratie en Inburgering, 2020]; COVID-19: Meertalige informatie over het coronavirus en de taalwetgeving, 2020]).

In the case of telephone contact tracing, the support for minority languages was implemented on a large scale. As contact tracing in Flanders served as one of the primary public modalities to sensibilize citizens of the dangers of COVID-19, and to inform and update them of current governmental measures in place, it was paramount that contact tracers not only reached speakers of the locally dominant, official language, but also minority language or foreign language speakers who may not be (sufficiently) proficient in Dutch. Since the project’s objective was to study the activity type of telephone contact tracing interaction in all its guises, it was necessary to include the telephone calls in languages other than Dutch in our data collection. In doing so, the researchers on the project were confronted with the observation and recording of contact tracing interactions in languages in which they were themselves not proficient.^[1] This also resulted in a delayed analytical process, due to the need for preparatory, third-party provided transcription and translation in order for the researchers to make sense of the data and carry out the analysis.

In this article, we explicitly topicalize our analytical experiences with COVID-19 contact tracing interactions in languages other than Dutch. Recent literature has stressed the importance of (i) transparency in dealing with multilingual challenges during the research process, including the relevance of spaces of linguistic non-understanding (LNU) specifically (van Hest and Jacobs 2022). Other literature has thematized the practical challenges generated by the COVID-19 pandemic when conducting research (Gailloux et al. 2022). In response to this literature, we address the specific challenge of embracing the epistemological implications of communicative distance between the researcher and the participants, which render real-time interpretation, intervention and analysis impossible. We also examine what reflexivity about such communicative distance between the researcher and the participants adds to our understanding of the institutional practices that form the central focus of this specific study, and our research project more generally. To do so, we present a Goffmanian analysis of the interactional data which were collected in languages not included in the researchers’ linguistic repertoires. Specifically, we use Goffman’s (1981) framework of participant roles to examine the contact tracers’ positioning in the interactions under scrutiny. In doing so, we hope to illustrate how engaging with the linguistic spaces of non-understanding created by the minority language data we collected crucially informs and adds nuance to our analytical understanding of the COVID-19 contact tracing call as an institutional activity type. In the next sections, we first outline Goffman’s (1981) framework and then discuss our methods for data collection and analysis. Section 4 presents our results, structured according to the different types of role shifting we observed in our multilingual data. We finish our article with a number of concluding remarks about the analytical value of gathering data in spaces of linguistic non-understanding.

2 Participant roles and positioning

Goffman’s (1981) outline of participant roles in terms of formats of language production and reception occurs in his essay on “footing” which develops the theme of how interactants are aligned to what is being said or exchanged. Alignments may be characteristic for the whole duration of the interaction, but they can also shift multiple times in the course of it. Code selection (i.e. choice of medium of interaction) and code switching are listed among the significant indicators of “(changes in) footing” (see also Slembrouck 2009). Goffman’s framework for the study of participant roles also entails that we attend to the “full physical arena” (1981: 136) and that we differentiate between participants who are centrally involved as key speakers and listeners, and participants who are situated on the side of the focal encounter, e.g. adopting a role as an (invisible) overhearer or an (actively co-present) observer. Evidently, the contract tracing interactions under scrutiny in this article fall within the former category of focused interactions: during a contact tracing call, a contact tracer [CT] and index patient [IP] are focused on one and the same activity, viz. discussing the patient’s recent COVID-19 infection. Nevertheless, these focused interactions take place with meaningful participation on the sides (e.g. while the CT talks to the IP, a call center team coach may be monitoring the phone call, or, in the case here, a researcher may be recording and listening in on the conversation). As communication develops along multiple complex paths of shared and not-so-shared foci of attention, it is not surprising that in focused interactions, there can be certain distractions from a central purpose, and changes in footing and shifts in participant roles may occur as a result. As will be discussed in our analysis, both observations apply to our data set.

Instead of simply discussing interaction in the straightforward terms of a speaker and hearer, Goffman (1981) delineates specific ‘production’ and ‘reception’ formats. Production roles that can be taken on by an interlocutor include (i) ‘animator’, (ii) ‘author’, and (iii) ‘principal’. Respectively, these pertain to those who (i) in effect produces the words, (ii) composes the words, and (iii) carries official or public responsibility for what is being produced. Reception roles, then, refer to whether one is being ‘addressed’ or not, i.e. ‘unaddressed’. Within the latter category, one can for example take on the role of an ‘eavesdropper’ or ‘overhearer’ as an unaddressed participant who is still part of the interactional space. We believe that a discussion of listening roles is crucial here, and this extends into roles which simply register what goes on (e.g. recording an interaction, taking notes), or roles which entail active coaching from the side (e.g. whisper instructions) or roles which are evaluative in nature (e.g. judge whether things are done well). As Scollon (2006) notes, the three categories of animation, authoring and evaluative responsibility which Goffman (1981) introduces have pendants at the level of reception roles.

Within the context of our study, and especially so for the data collected in minority languages, this Goffmanian framework^[2] can help provide salient insights into the contact tracing telephone conversation. First of all, contact tracers can in some respects be considered ‘animators’ for most of the conversation, as they enact the requests for information from a script with questions on their computer screen, which was authored by the Agency of Health Care, and was updated continuously during the pandemic (De Timmerman et al. 2023). Of course, their engagement in interaction entails more than just reading from the screen. This applies, for instance, when a specific pre-formulated question is staged in a particular way. A dual interpretation as animator and author is thus invited. Secondly, as the researcher-fieldworker is not able to participate directly in the contact tracing call, in the sense that the researcher could talk to the tracer but not to the IP, the contact tracers act as a spokesperson for the research project and an animator for the research-related questions about informed consent and the questions posed to the IP for specific sociological background variables (e.g. gender, age, etc.). As such, the researcher-fieldworker who listens in online while the telephone call is being conducted is typically an eavesdropper with the role of an observer who also relies on technological means to record the call (note that, although the index patients know the conversation is being recorded, they are typically not aware of the researcher’s co-presence). In calls which take place in Dutch, English or French, the researcher-fieldworker can follow the interaction as it unfolds. However, for contact tracing interactions in languages the researcher is not proficient in, the contact tracing interaction becomes a space of linguistic non-understanding in which the researcher’s role shifts from being an interested eavesdropper to a non-understanding overhearer who handles the recording equipment. It is precisely this complex universe of distributed participant roles, and the shifts in alignment which the use of particular languages can bring about, which we will focus on in this article. Before we delve deeper into how we see these participant roles manifested in our LNU data, and what this tells us about the interactional activity type that is the COVID-19 contact tracing telephone conversation, we will first cover the relevant information on the methods and data used in this study.

3 Methods and data

The data on which we report in this article were collected during the third phase of a one-year inter-university research project funded by the Flemish Research Council (FWO).^[3] The overall goal of the project was to (i) map the interactional practice of this emergent institutional activity type, and (ii) to help improve interactional practice by offering advice and designing training modules to help contact tracers carry out their institutional brief. The project was split up in three distinct phases, which corresponded with three separate research foci and data collection efforts. It is important to note how the data collection processes throughout the three phases were affected by the restrictive conditions of fieldwork and data collection which applied during the different stages of the pandemic.

In a first phase, our main goal was to develop an accurate and detailed picture of how contact tracing conversations effectively took shape as an existing interactional practice which was turned into a large-scale institutional enterprise overnight. Contact tracing was done through call centers, and, at the Flemish Agency of Health and Care’s request, a brief of patient-centered support was specifically added to that of contract tracing and symptom registration. In the first stage of the project, we recorded 100 contact tracing phone calls in Dutch over a period of two months. This data collection phase started in December 2020, when Flanders was in complete lockdown, and all non-essential social contact was strictly forbidden. Because we could not travel and visit the call center on-site, participant recruitment options were severely limited, and we were forced to rely on a call center representative to shortlist potentially interested contact tracers, all of whom were then asked to participate in the project via Microsoft Teams and the call center’s official email service. The audio recording of the contact tracing calls took place remotely, as calls were recorded using Microsoft Teams’ screen sharing function (with the researcher-fieldworker physically located at home). The contact tracers who participated in the project had given their written consent for the recordings beforehand via email, while the contacted citizens gave oral consent. They did so twice: once before the recording started, and once after the contact tracing conversation proper had ended. The method of remote participant recruitment and call recording via a digital exchange platform was maintained during the two project stages that followed. The results yielded by this first phase of the study demonstrated considerable functional diversity within the contact tracing conversation. We found that the ideal CT is able to balance several functions and roles simultaneously; (i) giving advice and providing information adequately and appropriately, (ii) requesting and receiving relevant and reliable information, (iii) maintaining a care-orientation and securing compliance, and (iv) safeguarding their status as a government representative and task-oriented expert (see Bafort et al. 2023; De Timmerman et al. 2023; Slembrouck et al. 2023). For our analysis in this paper, we build on these insights to investigate the interactional dynamics of CTs, IPs and the overhearing researcher in spaces of LNU.

The project’s second phase was dedicated to using the insights from the first phase to help improve the institutional contact tracing practice. To that end, we designed a training module based on our initial interactional analyses and conducted an efficacy measuring experiment with a pilot group and control group of contact tracers who respectively had and had not received the training. Because the government restrictions at the time were more lenient, non-essential mobility was at that point discouraged, but not prohibited. We were consequently able to organize the training module on-site. The training sessions were prepared online in larger MS Teams meetings, and given on site by the researcher-fieldworker, together with one of the call center’s in-house coaches. Four groups of tracers received the training over the course of four days. Each session was attended by a second member of the research team. After the sessions, a second set of 66 contact tracing calls in Dutch were recorded during April, May and June of 2021.

The third phase of the project focused on the dimension of patient diversity. It involved a comparison with local contact tracing done via home visits, in addition to telephone contact tracing conducted in languages other than Dutch. The latter set of data is the focus of this paper. Data were collected for the lingua franca languages, French and English, and two commonly used minority languages, Turkish and Arabic. The condition of remote recording via a digital platform continued to apply. Over the course of 10 months, from late April 2021 until early February 2022, 57 additional conversations were recorded and observed by the researcher-fieldworker: 25 in French, 12 in English, 14 in Turkish, and 6 in Arabic. The design also included Bulgarian, as a less commonly represented migrant language, but no calls in Bulgarian presented themselves. As is reflected in the lower number of recordings and the longer duration of the data collection, recording these calls proved to be much more time-consuming than was the case for the 166 Dutch calls of phases one and two, as the number of calls taking place in languages other than Dutch was significantly lower, and we had access to just a handful of tracers who were proficient in Arabic, Turkish (and Bulgarian).

As will be elaborated in the analysis section of this paper, the multilingual data collection of phase 3 posed particular practical and methodological limitations. Although projects increasingly hire bilingual research officers, as is often the case with qualitative (socio)linguistic research projects on minority languages, we relied on external language workers throughout the research process, particularly for the transcription and translation of the data. Such language workers often continue to be consulted during analysis for advice on the interpretation of particular formulations and interactional sequences. In addition, in the case of the multilingual data, obtaining the patient’s informed consent hinged even more on the participating contact tracers. Remember that, for all calls, the observing researcher-fieldworker was an overhearer/eavesdropper whose presence was neither audible nor visible to the index patient at the other end of the telephone line. This was the case across the project’s three phases. For the multilingual calls in Arabic and Turkish, this posed additional challenges of linguistic non-understanding, as the language choices affected the fieldworker’s capacity to act as an overhearing interactant (cf. supra).

Although the research process in the phase including data in languages other than Dutch proved significantly more difficult than in the first two phases, the intricacies of analyzing and interpreting the minority language data elicited interesting insights and arguably led to a more complete, nuanced view of the contact tracing call as an institutional activity type. When considering the participation framework, we specifically note salient shifts in participant roles throughout these interactions, which, as will be illustrated below, can be considered a particular affordance of the data being collected in a space of LNU. We believe these shifts (i) occur interactionally as real-time changes in footing enabled by an underlying context of linguistic non-understanding, and (ii) are institutionally meaningful precisely because they entail observable deviations from the prevailing CT-practice in the dominant language. We should note, however, that the condition of linguistic non-understanding not only affects the data collection process from the perspective of the researcher-overhearer, but that it may equally impact the focal interaction between the CT and IP. As will become clear from the excerpts presented in our analysis, CTs and IPs often actively manage their differing linguistic backgrounds through code-switching, humor, extralinguistic cues, etc. While we do highlight such examples in our analysis, we will predominantly focus on questions of researcher reflexivity in spaces of linguistic non-understanding.

4 Participant role shifting in spaces of linguistic non-understanding

In order to accommodate the multilingual aims of the Flemish government’s contact tracing endeavor, contact tracers with a minority L1 who expressed a willingness to conduct calls in these languages were effectively asked by the call center to operate in some respects as on-the-spot translators/interpreters. They needed to translate the Dutch script into a minority language and translate the answers received in a minority language into Dutch input for the data registration system, without receiving any formal training to conduct such a task. This was in addition to their existing roles as call center operatives with a para-medical brief (see Bafort et al. 2023; De Timmerman et al. 2023). Operating in a minority language thus arguably added to the complexities of the contact tracing activity type, and the contact tracer’s already manifold interactional responsibilities.

In one of the Turkish conversations which we recorded, for example, confusion arose when the contact tracer asked the index patient for their date of birth. The tracer, who is fluent in Turkish and has Turkish cultural heritage, confused the months December and January, and had to ask the IP: “December is the first month of the year, right?” The translator whom we employed to process the Turkish data commented on this rich point (Agar 2006; Sandel 2015) when transcribing this particular interaction, stressing that many Turkish-speaking people of immigrant descent often only know the names of the months in the local dominant language. It is worth highlighting here that the two translators whom we hired to process the minority language data differed in terms of their background and professional status. The Turkish data were translated by a qualified postgraduate in translation and interpreting, who was familiar with the project and its interactional sociolinguistic methodology, and hence provided clarifying comments like the one above without having been asked to do so. This was not the case for the Arabic data, which were processed by a professional translator who works for a translation agency.

These and other, similar examples of communicative friction are plentiful in our multilingual corpus of contact tracing data, and are largely in line with observations made by existing research on language use in spaces of non-understanding about the need for externally-provided cultural and specific linguistic expertise to understand the data fully (see, for example, Andrews et al. 2019; Androulakis 2013; Duran 2019; van Hest and Jacobs 2022). More unique to our corpus, however, is our observation of the salient shifting of participant roles which is arguably caused or afforded by the multilingual dynamics of the interaction. In the following subsections, we home in on three distinct types of interactional role shifting which emerged in our data, as a result of changed conditions of code: (i) the contact tracer’s role as a spokesperson for the research project and its research team; (ii) the tracer’s interactional balancing between the roles of institutional agent, cultural mediator, and patient-centered interlocutor, which may be afforded by the space of linguistic non-understanding and the related lack of practice-monitoring overhearers; and, finally, (iii) the shifts in role and interactional (in)visibility of the observing researcher-fieldworker. Note that the observed role dynamics and the associated shifts from the established roles only emerged at a late stage of data analysis, as this became only possible after the data had been transcribed and translated. These dynamics result in a picture that complements our understanding of the practices/interactional dynamics that were observed in the CT conversations in Dutch.

5 Discussion and concluding remarks

In summary, the six data excerpts we have discussed in this article aptly illustrate how the condition of linguistic non-understanding, which characterized the third phase of the research project and data collection of Turkish and Arabic data, at times influenced the roles of the interlocutors as participants in a COVID-19 contact tracing telephone call. The shifts occurred in various forms, e.g. (i) when the contact tracer’s role as an animator-mouthpiece for the research project temporarily shifted to reflect one of a more active and involved spokesperson, or (ii) when the overhearing researcher-fieldworker temporarily became more interactionally visible. Changes in the tracer’s stance towards government vaccination policy occurred as well, up to the point where contact tracers’ interactional roles and alignment shifted to express explicit affiliation with the religious-cultural background of the index patient. Interestingly, it is precisely the condition of linguistic non-understanding, which characterized the collection of these specific data, that turned out to be the source and instigator for a number of shifts in the participants’ alignments.

A second conclusion we can draw on the basis of the discussed excerpts bears on how practices of participant alignment need to be understood exactly. Goffman’s taxonomy of participant roles may suggest that this is all about technical aspects of doing speaking and listening. Instead, we wish to stress that an analysis of “footing” should not be separated from the analytical realm of interactional meaning-making more generally, in the sense that more is at stake here than just communicative roles which can be interpreted in ways devoid of expressed content or displayed attitude. We noted how in one instance the CT’s apparently technical explanation about the reasons for collecting a particular set of language-specific data became a language-ideologically entrenched statement about the social value of particular languages, precisely as a result of the space of linguistic non-understanding which was present at the time. Thus, slippage emerges here from reflections on one’s interactional role, to value-laden positionings on the status of the languages involved. In a similar vein and turning to another aspect of our analysis, one can raise the question: at which point does the interactional imperative of patient-centeredness in a framework of contact tracing begin to overlap with a framework of religious-cultural affiliation, especially when the call takes place in a minority language? The dynamics of the latter manifested itself under changed conditions of (in)ability to overhear in the linguistic codes that are used to conduct the contact tracing call. Indeed, similar to how Irvine (1996) draws our attention to “intricate laminations of participant roles” (p. 157), overlapping and coalescing roles emerged in our data as well. To illuminate their complexity, we arguably need an “analytical focus on intersecting frames and dialogical relationships” (Rehnberg 2023: 4), and it is precisely changes in footing which appear to play a major role here and bring these role laminations to the fore.

A third and final point in our conclusion is that methodological reflexivity about the implications of spaces of linguistic non-understanding crucially informed insights which have relevance for the scope and nature of the project’s overarching research questions. Without engaging with the spaces of linguistic non-understanding presented by our Turkish and Arabic data, we would have been left unaware of the interactional role shifting that may occur when institutional encounters take place in minority languages. Indeed, we can conclude that the inclusion of contact tracing calls in languages other than Dutch in our attempt to exhaustively document and understand the COVID-19 contact tracing call as a distinct activity type has been a necessary requirement which allowed us to engage with intricate positioning by the contact tracers in carrying out their institutional tasks. Without these minority language data – despite the difficulties faced when collecting, processing, and analyzing them – our view of the institutional practice under scrutiny would have arguably been incomplete at its best, and inaccurate at its worst.