Institutional CA

As indicated in the Introduction, this article adopts the perspective of institutional CA. Institutional CA, a branch of CA which emerged in the late 1970s, can generally be described as using CA methods in order ‘to examine the operation of social institutions in talk’ (Heritage Reference Heritage, Kristine and Robert2005:104). More specifically, scholars (e.g. Schegloff Reference Schegloff, Drew and Heritage1992; Heritage Reference Heritage, Kristine and Robert2005; Heritage & Clayman Reference Heritage and Clayman2010; Arminen Reference Arminen2016) recognise three research objectives to institutional CA:

1. a. Explicating interactional practices by referring to their institutional contexts;

2. b. Understanding the workings of institutions by focusing on their interactional dimension;

3. c. Contributing to wider issues ‘that are beyond the talk’ (Heritage Reference Heritage, Kristine and Robert2005:115)

This article contributes to the first objective. Examples of previous work under this category include Button (Reference Button1987), Stoke & Edwards (Reference Stokoe, Edwards, Alice and Ehrlich2009), and Gafaranga (Reference Gafaranga2017b). Button has investigated the ‘logic’ of the features of sequence organisation he had observed in job interviews, Stoke & Edwards have addressed the ‘raison d’être’ of ‘ostensibly silly questions’ in police-suspect interviews, and Gafaranga has developed an account of the diverse uses of ‘how are you?’ in general practice consultations in the UK.

Studies that aim to elucidate interactional practices referring to their institutional contexts share three key methodological requirements. First, they adopt a ‘comparative approach’ (Drew & Heritage Reference Drew, Heritage, Drew and Heritage1992; Arminen Reference Arminen2016). They start by observing similarities and differences (if any) between ordinary conversation and institutional talk and proceed to account for observed differences in terms of the institution’s ‘fingerprint’ (Drew & Heritage Reference Drew, Heritage, Drew and Heritage1992). This is important because, if there are no differences, there is no need to invoke the institution. A second important requirement in these studies is to demonstrate, ‘from the details of talk or other conduct in the materials that (one is) analyzing’ (Schegloff Reference Schegloff, Drew and Heritage1992:110), the ‘relevance’ for participants of the claimed explanatory aspect(s) of the institution. This demonstration is necessary because, without it, there is a risk of falling into the ‘positivistic stance, even though the animating concerns may be drawn from quite anti-positivistic theoretical sources and commitments’ (1992:102), Third, there is a need to demonstrate the ‘procedural consequentiality’ (Schegloff Reference Schegloff, Drew and Heritage1992:111) of the claimed institution to the talk. Analysts must demonstrate that the fact that talk is conducted in a specific institution has consequences ‘for its shape, form, trajectory, content, or character’ (1992:111). In turn, this requirement is based on the theoretical position that ‘context is talk shaping’ (Drew & Heritage Reference Drew, Heritage, Drew and Heritage1992; Heritage & Clayman Reference Heritage and Clayman2010). I have attended to the first methodological requirement in the Context of literature section above and I attend to the remaining two in the Analysis section below.

The broadcast news interview as institutional talk

According to Drew & Heritage (Reference Drew, Heritage, Drew and Heritage1992), institutional talk broadly divides into two categories: ‘formal’ and ‘non-formal’ institutional talk. Non-formal institutional talk takes place in ‘private rather than public contexts’ and adopts ‘turn-taking procedures (which) approximate conversational or at least “quasi-conversational” modes’ (Drew & Heritage Reference Drew, Heritage, Drew and Heritage1992:28). Formal institutional talk, by contrast, often adopts specialised turn-taking systems, including the possibility of a ‘pre-allocation of turn types’ (Drew & Heritage Reference Drew, Heritage, Drew and Heritage1992:26). Also, formal institutional talk is often designed for an overhearing audience. Examples of formal institutional settings include courtrooms, classrooms and, most importantly for this article, broadcast news interviews.

Clayman & Heritage (Reference Clayman2002:7) describe the broadcast news interview as ‘a familiar and readily recognizable genre … [but whose] members share a loose family resemblance rather than a rigid set of defining attributes’. As a genre, the broadcast news interview comprises four main subgenres, depending on the interviewees’ identities, namely accountability interview (interviewee: public office holder), expert interview (interviewee: somebody with specialist knowledge of a relevant topic), experiential interview (interviewee: somebody with first-hand experience of a relevant event), and affiliative interview (interviewee: other media professionals, e.g. correspondents) (Montgomety Reference Montgomery2008). Interviews may also be described as one-to-one interview or as panel interview depending on the number of interviewees. However, Clayman cautions against reifying ‘these categories or (treating) them as wholly determined by the ‘objective’ nature of the IEs’ identities. Specific interviews may straddle the boundaries between genres’ (2014:653). Two of these categories are present in my corpus, namely accountability interview and expert interview.

As we have seen, the broadcast news interview is a prototypical example of formal institutional talk. As such, it has a specialized turn-taking system whereby opportunities for action are differentially distributed between/among participants. A simple rule organises turn-taking in news interviews: ‘Interviewers restrict themselves to questioning and interviewees restrict themselves to answering interviewer’s questions, or at least responding to interviewer’s questions’ (Clayman & Heritage Reference Clayman2002:97). More importantly for the concerns of this article is the fact that, in news interviews, talk is designed for an overhearing audience. Research has identified a variety of interactional practices through which interview participants reveal their orientation to this aspect of talk. These include: (i) ‘avoiding receipt tokens during IEs’ elaborated turns’, (ii) ‘displaying understanding for the audience’, and (iii) ‘explicit references to the audience and full-form IE address terms within closings’ (Clayman Reference Clayman, Sidnell and Stivers2014:636–37). This article argues that, in broadcast news interviews in Rwanda, other-language recalibration is one of the strategies through which interview participants orient to the overhearing audience. In developing this argument and following Schegloff’s recommendations, in the next section, I begin by demonstrating the relevance of the overhearing audience for interview participants.

Orienting to the overhearing audience

According to Schegloff (Reference Schegloff, Drew and Heritage1992:110), the relevance of an (aspect of) institution must be demonstrated in ‘the details of talk or other conduct in the materials’ one is analysing. One conduct through which interview participants reveal their orientation to the overhearing audience is, according to Clayman, ‘explicit reference to the audience’ (2014:637). In news interviews in Rwanda, it is common practice for participants to explicitly refer to the audience. Here are some examples. Extract (7) comes from an interview with members of a political party (Liberal Party) which is celebrating its thirtieth anniversary. IR has asked IE, a former leader of the party, a question about the legacy of the party. In line 2, IE mentions the youth as the real target of his message (nabwira urubyiruko…).

Extract (8) comes from a panel interview with known panafricanists on the status of Rwanda on the international scene. IR asks IE what he would tell the overhearing audience (ikintu wabwira abatwumva) about this status in a few words (line 4).

Also consider extract (9), which comes from a panel interview on the eve of the referendum that removed the presidential term-limit clause from the Constitution (2003/2015).

In lines 1–3, IR tells the interviewees that they have made the discussion very technical in terms of legal procedures and goes on to formulate IE’s contribution and, in lines 4 and 5, he says in so many words that the interviewee’s talk had been addressed to those “who are following this programme”.

Since, as this evidence shows, participants are aware of the overhearing audience as the real recipients of their talk, we can anticipate more or less safely that this orientation to the audience has repercussions on the organisation of talk and is consequential to it. As Heritage & Greatbatch (Reference Heritage, Greatbatch, Boden and Don1991:109) put it, ‘the footing of news interviews as talk oriented towards the overhearing audience is managed at all points over the course of the talk and not merely at these points where an overt reference to the audience or some other specially audience-directed activity takes place’. One point where this orientation is managed, I want to argue, is in the specific organisation of other-language recalibration repair.

Formulating other-language recalibration repair as talk directed to the overhearing audience

Garfinkel & Sacks (Reference Garfinkel, Sacks, John and Edward1970:351) use the term formulating to refer to ‘conversationalists’ practices of saying-in-so-many-words-what-we-are-doing’, of ‘explicitly naming the kind of action one is doing’ (Fele Reference Fele2023:84). According to Depermann (Reference Depermann2011:117), ‘formulations are a principal means participants use to display and secure that their interactions are self-explicative, intelligible and accountable’. Any aspect of talk, not just the actions accomplished, can be formulated (Garfinkel & Saks Reference Garfinkel, Sacks, John and Edward1970:351). In the news interviews in the corpus, participants sometimes formulate the status of other-language recalibration as talk addressed to the overhearing audience. That is to say, participants may, in addition to recalibrating their talk, say in so many words that their action is for the benefit of the overhearing audience. Consider extract (10). A high-ranking government official has been asked to comment on media reports that the ruling party stands accused of atrocities during the civil war era (1990–1994). The official reacts saying that these allegations are scandalous.

In the extract, IE recalibrates a Kinyarwanda item (agahomamunwa, line 4) using its French equivalent (scandale, line 5) and, in addition, specifies that he does it “for those (of the audience) who speak French”. A second case is extract (11). Participants are debating the situation of somebody who is appearing in court in Rwanda, appealing a sentence he has been given for terrorism charges. The defendant has changed his defence strategy, moving from an earlier guilty plea to the denial of guilt. Because of this, participants are saying that, now, he could also be accused of obstructing justice. IE uses the English phrase “obstruction of justice” (line 2). IR thinks this expression could be challenging (amagambo akomeye, line 5) for the audience and asks IE to translate, that is, to recalibrate it for the audience (wayasemurira abantu, line 7) and IE complies in subsequent talk (not shown).

As Garfinkel & Sacks (Reference Garfinkel, Sacks, John and Edward1970:353) say, in talk-in-interaction, formulating is rather infrequent. However, despite its scarcity, formulating is generally agreed to be one of the most powerful sources of evidence in CA (Sidnell Reference Sidnell, Sidnell and Stivers2014:81–82). On this basis, we can proceed with confidence that, in news interviews in Rwanda, other-language recalibration is a device for relating to the overhearing audience even when it is not formulated as such.

The procedural consequentiality of audience-directed talk

So far, I have demonstrated that participants in news interviews in Rwanda orient to the overhearing audience explicitly referring to them, among other practices. And I have provided evidence that other-language recalibration is done for the benefit of the overhearing audience. In the data, this function is sometimes explicitly formulated by participants themselves. I now turn to Schegloff’s (Reference Schegloff, Drew and Heritage1992) second recommendation regarding the study of institutional talk, namely procedural consequentiality. Specifically, in this section, I show that the fact that other-language recalibration repair is talk addressed to the overhearing audience shapes the organisation of the repair practice.

The absence of other-initiated self-repair. As we have already seen, a key structural feature of other-language recalibration repair is the absence of the other-initiated self-repair format. Typically, interlocutors do not initiate self-repair on each other’s talk. A rare exception, which in my view confirms the norm, is extract (11) above. In the extract, IR invites IE to self-repair (“can you translate them for people?”, line 7) and provides an account for this invitation (“you’ve used difficult words”, line 5). This account not only points to the ‘dispreferred’ nature of other-initiation in this context (Pomerantz Reference Pomerantz, Atkinson and Heritage1984), but also suggests that IR initiates repair because he is unable to accomplish it himself (see other-initiated other-repair below). Since other-language recalibration is done for the benefit of a third party, namely the overhearing audience, it is very easy to see why other-initiated self-repair is rare, if not impossible. Other-initiation of self-repair in other-language recalibration is avoided for exactly the same reason that receipt tokens are avoided in news interviews in general. In news interviews, receipt tokens are avoided because their use would make the situation look as if co-present participants were the real recipients of the talk (Clayman Reference Clayman, Sidnell and Stivers2014:636). Just like the use of receipt tokens, other-initiation of repair is a recipient’s action. On receiving an interlocutor’s talk, the recipient either produces the next relevant action or they initiate repair on it. In other words, in news interviews, participants avoid initiating other-language recalibration repair in the format other-initiated self-repair on interlocutors’ talk in order to avoid being seen as the recipients of that talk.

Respecifying other-initiated other-repair. We have also seen that, in other-language recalibration, the format other-initiated other-repair is common. In passing, we can note that repair in this format is always initiated and executed by the interviewer (see extracts (2), (12)–(14)). If other-language recalibration is seen as talk directed to the overhearing audience and that the fact that it is always initiated and executed by the interviewer is taken into consideration, an account of this repair format turns out to be straightforward. The format other-initiated other-repair in other-language recalibration can be likened to what Clayman (Reference Clayman, Sidnell and Stivers2014: 636–37) terms ‘displaying understanding for the audience’. To do this, IR ‘simply provides an adequate reformulation of the interviewee’s previously stated position’ (Heritage Reference Heritage and Dijk1985:108). In the case at hand, IR simply changes the language of IE’s talk. More importantly, we have already seen that broadcast news interviews have a specialised turn-taking system characterised, among other things, by ‘action pre-allocation’ (Clayman & Heritage Reference Clayman2002:37). The interviewer asks questions and the interviewee answers them (Clayman Reference Clayman, Sidnell and Stivers2014:631). It would appear that the same differential distribution of interactional actions obtains at the level of other-language recalibration. The interviewer can initiate and execute repair on interviewees’ talk, but interviewees can only initiate and execute repair on their own talk.

Heritage (Reference Heritage and Dijk1985:108) identifies a three-step sequence over which IRs display understanding of IEs’ talk for the overhearing audience: statement + reformulation + confirmation. This three-step structure is evident in extract (2) and is also found in extract (12). Participants are talking about inter-Africa trade. IE notes that, for it to be possible, cargo aircrafts have to be bought. In the extract, the item indege z’amacargo in line 1 is recalibrated as indege z’imizigo in line 5 and the repair is confirmed by repeating the proposed repair solution in line 6.

Even more interesting is extract (13). An MP from the ruling party is saying that, except for economic development, everything is ready for the country to take off. The recalibration of the item “economic development” in line 6 is undertaken at line 8 and the recalibration is confirmed at line 9.

However, on a second look, the account of this repair format turns out to be not as straightforward as it first appears. Two key questions can be asked. First, if IE’s talk is addressed to the overhearing audience, how come IR initiates and effects repair on it? Other-initiation of repair is the preserve of the talk recipient (see above). In addition, according to Schegloff and colleagues (1977:380), ‘other-correction’ is impossible in principle. However, Schegloff and colleagues (1977:380) also note that there may be environments, such as storytelling, where participants act as a ‘team’ and, as such, can correct each other overtly. This idea of talking teams is further developed in Lerner (Reference Lerner1993:237) who shows that talking teams may consist either of ‘enduring social units (e.g. couples)’ or of ‘momentary associations’. Either way, these teams are talked into being in a variety of ways. Crucially, as Lerner says, in a team, ‘one party (may) speak on behalf of a co-participant or report on their speech and action’ (1993:230). Interestingly, other-correction is given as a good example of an action which can be done on behalf of a co-participant. Bolden (Reference Bolden2013, Reference Bolden2024) takes the idea of talking teams even further and shows that one particularly fertile environment for team formation is when talk is addressed to third parties. In such circumstances, a participant may operate a correction on behalf of a co-member and do so ‘for their recipients’ (Lerner Reference Lerner1993:230). Thus, according to Bolden (Reference Bolden2024:211), ‘within-team correction might be considered as a form of repair that is closer to self-correction than other-correction’. It is along these lines that the question above can be answered. IR claims co-membership with IE and produces a recalibration repair on his/her behalf for the benefit of the overhearing audience.

Second, as the data shows, the so-called confirming step consists overwhelmingly of repeats of previous turns (e.g. extract (2), (12), and (13)). A rare exception can be found in extract (14). IR asks IE why some people (himself included) move from allegiance to the current government to allegiance to the opposition and the other way round. In answering the question, IE starts by quoting a famous musician from Ivory Coast who is reported to have said that only stupid people don’t change.

In quoting the singer (lines 1 and 2), IE uses French. In lines 3 and 4, IR recalibrates the quote using Kinyarwanda and, in line 5, IE ratifies the recalibration. Exceptionally, here, IE does not repeat IR’s prior talk. Instead, he uses the emphatic French agreement token voilà.

Research has identified two main types of repeat in talk-in-interaction, namely repair initiating repeat and registering repeat (Schegloff Reference Schegloff1997; Persson Reference Persson2015; Rossi Reference Rossi2020; etc.). The function of a registering repeat is to register receipt ‘of the preceding turn and the actions done in it’ (Schegloff Reference Schegloff1997:527). That is to say, through registering repeat participants do being the recipients of the talk in the prior turn. Repeat in other-language recalibration seems to be of this second category as it does not initiate repair on prior talk. However, there is a problem. Since, in all other aspects, interview participants avoid recipient actions (e.g. avoidance of receipt tokens, Clayman Reference Clayman, Sidnell and Stivers2014:636; avoidance of other-initiated self-repair (see above)), why would they do otherwise here? Why would interview participants register receipt of talk (other-language recalibration in the preceding turn) which is not addressed to them?

Arano (Reference Arano2018) has focused on what he/she calls ‘post-other-correction repeat’ in multilingual interaction between native and non-native speakers. They identified a three-step structure consisting of self-initiation + corrective demonstration + third position self-correction. As it can be seen, the structure is similar to that of other-correction as described in Schegloff and colleagues (1977:378–79), but it is also different in very important ways. In their work, the repair solution is a ‘candidate solution’, is ‘low on the confidence/certainty scale’, and is ‘proffered for confirmation/rejection’. In Arano’s study, as the label indicates, step two is high on the certainty scale and is a ‘demonstration of correct behaviour’. And this is so because of the underlying asymmetry of expertise/knowledge between participants. As a result, while, in Schegloff and colleagues (1977), the third step is a confirmation/rejection, in Arano, it is a ‘third turn self-correction’. Arano suggests that the same structure can be found in other settings characterised by asymmetries, including workplace settings (2018:17). One can therefore argue that, in other-language recalibration in news interviews, a similar process is at work. IE overhears IR’s recalibration, uses it as a ‘corrective demonstration’ and self-repairs in turn three. To put it simply, IE claims the recalibration repair for him/herself and deploys it for the overhearing audience. Thus, the format of other-language recalibration previously described as other-initiated other-repair can be re-analysed as two imbricated self-initiated self-repairs, one by IR (claiming membership in a talking team) and one by IE (after a corrective demonstration), both undertaken for the benefit of the overhearing audience.

Respecifyng self-initiated other-repair. If other-language recalibration is seen as talk addressed to the overhearing audience, the format we have previously analysed as self-initiated other-repair too needs to be reconsidered. As we have seen, typically, self-initiation of other-repair is done indirectly through ‘displays of trouble’ (Reichert & Liebscher Reference Reichert and Liebscher2012), for example, ‘other-languageness formulation’ (Gafaranga Reference Gafaranga2015). This can be seen in extract (5). A further example of the use of other-languageness formulation is extract (15). Participants are talking about the tendency Rwandan parents have of sending their children abroad for better opportunities.

At line 1, IR describes the parents’ tendency using the French noun phrase mentalité and, immediately after, he indicates its other-languageness (mu gifaransa niko navuga wenda, line 2). Subsequently, IR provides the recalibration (imyumvire, line 4).

While there is evidence that the display of trouble by one participant can make a recalibration by the interlocutor relevant, the issue is whether it makes other-language recalibration by next speaker ‘conditionally relevant’ (Schegloff 1968). This is important because the latter possibility would be incompatible with the view that other-language recalibration is talk addressed to the overhearing audience. A closer observation of the data reveals that displays of trouble do not make other-language recalibration by an interlocutor conditionally relevant. In extract (5), at line 2, IR displays trouble, stops speaking and, immediately after, IE provides a recalibration. In extract (6), by contrast, at line 4, IE1 produces an other-languageness formulation but no recalibration follows. A similar situation occurs in extract (16) below.

Between these two extremes, we find situations such as the one in extract (17) where IE displays trouble (line 4) and goes on to provide the recalibration himself (line 6).

There are also situations such as extract (15) where both participants orient to producing a recalibration and end up producing the same recalibration in overlap (lines 3 and 4). All this suggests that, in other-language recalibration repair, the format initially thought to be a case of self-initiated other-repair is analysable as word search repair (Schegloff et al. Reference Schegloff, Jefferson and Sacks1977; Brouwer Reference Brouwer2003; Dressel Reference Dressel2020; etc.) and, within it, displays of trouble, not as initiating actions, but ‘as accounts … for not having produced’ a recalibration (Brouwer Reference Brouwer2003:538). Under this view, the format self-initiated other-repair is only apparent. As in the case of what initially appeared to be other-initiated other-repair (see above), the actual structure is that of two imbricated self-initiated self-repairs, the only difference being that, here, action-type pre-allocation does not obtain.

Summary and conclusion

Adopting a comparative perspective, this article opened by noticing organisational differences between repair in ordinary conversation and other-language recalibration repair in broadcast news interviews in Rwanda. It was observed that, while repair in everyday conversation can take any of four formats (self-initiated self-repair, self-initiated other-repair, other-initiated self-repair, and other-initiated other-repair), other-language recalibration repair can only take three of these. Therefore, the article set itself the aim of accounting for the observed variance, that is, of investigating the ‘logic’ of this structural reduction. Based on available literature according to which repair can be deployed, not to solve any problems of production/reception of messages, but ‘in the service of the action the speaker means to be doing with the talk’ (Kitzinger Reference Kitzinger, Sidnell and Stivers2014:242), I took as my working hypothesis that, in broadcast news interviews in Rwanda, other-language recalibration repair is used as a strategy for relating to the overhearing audience and that, in turn, this interactional use of repair is consequential to its shape. Analysis of the data confirmed this hypothesis. It confirmed the hypothesised interactional use of other-language recalibration repair and its consequence for the shape of the interactional practice. To be sure, analysis revealed that the structure of other-language recalibration is in fact even more reduced than it had originally been thought to be. The structure of other-language recalibration repair consists, not of three formats as initially thought, but of only one, namely self-initiated self-repair.

This finding advances our understanding of repair in talk-in-interaction, of the relationship between talk and its institutional context, and of talk in news interviews, albeit in a small way. But it also calls for further research. There is a need to know whether other-language recalibration repair is specific to the particular sociolinguistic context of Rwanda or whether it is a feature of bilingual contexts generally. If other-language recalibration is found in other sociolinguistic contexts, there is a need to know whether it adopts the same structure or whether its organisation is context-sensitive. More importantly, there is a need to investigate the structure of other-language recalibration repair in ordinary conversation. In this article, I have taken as a ‘benchmark’ (Heritage Reference Heritage and Silverman1997) the organisation of repair in conversation in general. Clearly a more reliable benchmark would have been the structure of other-language recalibration repair in everyday conversation, had it been available.