On second language/nonnative speakerism in conversation analysis: A study of emic orientations to language in multilingual/lingua franca couple interactions

While conversation analytic research on second language (L2) interaction tends to use categories such as L2/nonnative speaker and learner, these labels are not always rooted in the participants' emic orientations. This paper argues that the emic principle of conversation analysis should reverberate in the labelling of data types. It suggests that where overt orientations to nonnativeness are not pertinent, analysts should refrain from using stiff categorizations that excessively simplify speaker identities. Instead, participant labelling should be delayed until after emic analysis. Contrary to often explored institutional interactions, the current paper examines naturally occurring conversations between interlinguistic couples who could be described as users of English as a lingua franca or multilingual speakers, depending on the perspective. It explores the prevalence of sequential orientation to linguistic form compared to overall units of talk, to what degree language issues are repaired, and whether participants orient to language as novices/experts. The results indicate very little attention to language overall and where observable, interlocutors assume fluctuating speaker identities. Most repair work is unrelated to language, and displays of L2 identity (Kurhila, 2004) and expert/novice orientations (Brouwer, 2003) are rare. The interlocutors routinely produce a sequentially relevant next turn rather than correct nonstandard features. © 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

acquired through interaction, and used resourcefully, contingently, and contextually" (p. 296, emphasis removed). However, while the contributions made visible other, institutional roles over speakership identities, they failed to sufficiently address whether contrasting NNS/L2 interaction to NS/L1 interaction is a meaningful comparison in the first place.
At the same time, more research on English as a lingua franca (ELF) began to emerge. Here the focus was on English used as the global contact language among speakers from varied linguistic backgrounds. In the early days, the paradigm concentrated on NNS-NNS interactions and excluded NSs of English. More recently, the field has opened for reconceptualizations that both extend to include all speakers who use English as a contact language with anyone of a different L1 background (Seidlhofer, 2011), and involve translanguaging and the use of languages other than English under the umbrella of ELF (Jenkins, 2015). Although ELF has primarily been a (socio-pragma-pedagogical) linguistic enterprise, the analysis of ELF interactions has also interested conversation analysts, not least due to the global access which ELF provides to intercultural settings of various kinds. However, although conversation analysis (hereafter CA) is one of the main methodologies in ELF research (Firth, 2012;Kaur, 2016), only few contributions have been made toward CA from this line of work (e.g., Pietik€ ainen, 2018b;Santner-Wolfartsberger, 2015). 1 Notable exceptions include Firth's earlier work (e.g., Firth, 1996;Firth and Wagner, 1997) in which he argued that albeit from the surface level, lingua franca communication may seem disorderly, it is analyzable from the emic perspective of the participants and should therefore be included in the array of CA's research interests. The current article argues that investigating speaker orientations in translingual/lingua franca data allows for the scrutiny of speaker identities on a much deeper level than has been hitherto explored and grants the reevaluation of naming practices used for labeling participants in CA-SLA research.
CA-SLA work has mainly focused on institutional interactions, while the current paper studies interactions from one of the most ordinary non-institutional contexts: between married couples. It focuses on how couples who use English as their main couple lingua franca (Kl€ otzl, 2015; orient to language and particularly, what kinds of "troubles" they repair in their interaction. This will reveal details of their speaker identities and the underlying normativity that concerns language. Depending on the point of view, the interactions scrutinized can be described as L2 or ELF talk e and one of the main objectives of the paper is in fact to discuss which label would best suit to describe the interactions when investigated from an emic perspective. The paper is organized as follows: it first outlines relevant key concepts to the discussion concerning the emic perspective of CA, repair organization, and CA's labelling categories. In 2.2 the labelling categories of ELF research are outlined, after which I describe relevant background in CA-SLA research, contrasting it with findings from ELF in 2.3, and outline the research questions. Section 3 describes the data used in the study and section 4 outlines the methodology. A detailed analysis with examples in presented in section 5, after which the findings will be discussed and concluded in section 6.

CA's analytical principles
One of the fundamental premises of CA is that the phenomenon studied should be such that interactants demonstrably orient to it in the course of the interaction (Psathas, 1995:46). Participant orientation is visible through interactional organization: turn-taking and sequential organization, repair work, and preference organization (Seedhouse, 2005:252e253). Contrary to many other methods of linguistic/social research, CA does not approach research phenomena from researcherinitiated perspectives. It analyses the ways in which participants utilize phenomena "within specific systems of action" (Goodwin, 1984:243). This analytic principle is known as the emic perspective (Pike, 1966). The specific systems of action that speakers adhere to are thought to be simultaneously context-relevant and context-free: although speakers achieve social action through context-sensitive implementation e for example, linguistic forms "are used to embody and express subtle differences in social actions with social consequences" (Seedhouse, 2005:252) e the context-free machinery through which social world is talked into being operates on the level of interactional structure. This interactional organization has been the major interest of CA traditionally, and although current CA research has various context-specific directions of interest, analyses are performed in line with and reflected against the context-free, structural organization of interaction.

Participant orientation and repair
As an overarching analytic principle of CA, the emic perspective is also relevant when conversation analysts examine orientations to language. In its essence, an emic orientation signifies that normativity is considered endogenous, i.e., what participants themselves orient to as "correct" and "acceptable" should be perceived by the researcher as such in the context. Exogenous norms such as language standards and their relevance to the speakers can be reflected upon in the analysis, but they should not be used to evaluate or judge the speakers' performance where there is no endemic focus on such "correctness" in the data. Participants' emic orientations to language are most readily analyzable in repair organization: what is considered a repairable, who initiates repair, and how it is achieved. Schegloff (1987:210) defines repair as follows: By "repair" we refer to efforts to deal with trouble in speaking, hearing, or understanding talk in interaction. "Trouble" includes such occurrences as misarticulations, malapropisms, use of a "wrong" word, unavailability of a word when needed, failure to hear or to be heard, trouble on the part of the recipient in understanding, incorrect understandings by recipients, and various others. Because anything in talk can be a source of trouble, everything in conversation is, in principle, "repairable." There is an overwhelming prevalence of self-repair over other-repair (Schegloff et al., 1977). Kaur (2020) provides a condensed description of self-and other-initiated repair, and there is no need to repeat it here. Instead, I will outline Macbeth's (2004) distinction between repair and correction because it is useful in understanding language-related repair. Macbeth argues that although correction is a kind of repair, it has a special function. The premise of correction is that some type of an error has occurred e in a language classroom these errors commonly relate to language. In classrooms, where speaker roles are disparate, correction is usually performed in three turns: the first turn is initiated by the teacher and often takes the form of a question. The second turn is an answer from a student. The sequence-closing third is the teacher's remarks of the correctness of the student's turn or an evaluation of it. Correction can thus be seen as tied to the activity of learning and the achievement of classroom teaching. Although classroom correction also shows preference for self-correction, it is routinely initiated by the teacher, much more so than in naturally occurring interaction outside the classroom (Macbeth, 2004). The teacher's identity as a teacher can thus be displayed in their frequent initiations of correction cycles, whereas in contexts where speaker identities are in principle equal, one must have a special motive such as disagreement to make a correction instead of a sequentially appropriate next turn (Schegloff et al., 1977).

Categorization and naming practices in CA
In this section, the prevalence of the emic perspective in CA's labelling practices is discussed. Although CA claims that it resists "'obviously relevant' contextual features such as participants' social status, gender, race, etc." (Seedhouse, 2005:255), these are commonly presented as background information for readers (albeit sometimes implicitly and instead of race, participants' assumed L1 is often given). Many CA-SLA researchers collect data from speakers from language backgrounds of special interest, and already at this stage name the participants. Such naming categories can be, for example, "NSeNNS conversation" and "NNS classroom discourse" (Wong, 2005:160), "second language conversations" (Brouwer et al., 2004:75), or "English as a lingua franca data" (Santner-Wolfartsberger, 2015:254). Although CA positions that emic analyses should "show which of these innumerable, potentially relevant characteristics are actually procedurally relevant to those participants at [any] moment" (Seedhouse, 2005:255), studies that do just this are in the minority in CA-SLA literature. 2 To solve this problem, some more ethnomethodologically oriented CA scholars propose that alongside sequential organization, analysts should to a larger degree focus on the processes of categorization as the members' production (Housley and Fitzgerald, 2002) and do this "over time and across situations" (Day, 2008:995), because categories which are relevant for participants may not be demonstrable in each of their activities. For CA-SLA analysts, a thorough membership categorization analysis of the entire data set might seem obsolete, but at least refraining from etic speaker categorization before the data analysis has been done should be possible. It would help prevent labelling practices that do not rely on participant orientation and the unwelcome etic influence that such labels may carry with them upon the analysis, as will be argued later.
CA studies on so-called "L2 interaction" often scrutinize interactions between a L1 speaker/NS and a L2 speaker/NNS. The speakers' practices in these interactions are compared to each other, resulting in a description of "L2 talk" (see section 2.3). It is reasonable to ask whether NSeNNS interaction can reliably be used for this purpose, and why does the NS continue to be some kind on an exemplary, when achieving intersubjectivity requires efforts from both participants? In their critique of SLA, Firth and Wagner (1997) condemn this widely applied division between NS and NNS. They argue that "NNSs are unproblematically viewed as the NS's subordinates, with regard to communicative competence (the negative connotation of the 'non-' prefix is hardly coincidental)" (1997:291). Firth (1996) encouraged the use of the term lingua franca interaction, as the term emphasizes the legitimacy of the speakers as language users rather than as learners who are "conceived a priori to be the possessor of incomplete or deficient communicative competence" compared to the idealized NS (p. 241). In more recent CA-SLA literature, the term "L2 speaker" has come to replace NNS, but the effect seems mostly cosmetic. Just as the 'non-' prefix is problematic for NNS, 'second' is arguably subordinate to 'first' and is therefore equally problematic. It supposes that the speaker has a first language that is surmised to be better, more fluent, 'native' or 'mother tongue', and that the second language(s) can never reach as important status as the first. As the mere idea of using numerals to refer to languages has been widely criticized by contemporary literature on multilingualism, translanguaging, and multi-competence (see e.g., Canagarajah, 2007;Cook, 2016), it is hard to see why CA-SLA should continue to hold on to its "S" so tightly. Cook (2016) proposes that "multilingual" would be the closest term to replace "L2 user" because it does not presuppose high proficiency or exclude proficiency in multiple languages (ibid:5). ELF research, on the other hand, avoids the problem by using the terms lingua franca interaction and lingua franca speaker/user. The next section briefly reviews the principles which have guided these practices in ELF research.

How ELF views speakers
The research paradigm of ELF has develop over the last 20þ years to challenge the hegemony of native speakerism in English and to give a voice to multilingual users of English in their own right. The field is empirically driven: its major principle has been to investigate actual (primarily spoken) interaction in its naturally occurring environments and to describe what speakers do and who they are from an endonormative standpoint. In this respect, ELF and CA are undeniably compatible. Contrary to earlier definitions, 3 current ELF research attempts to move away from describing speakers from the perspective of what they are not and instead emphasizes the choice of medium, defining ELF as "any use of English among speakers of different first languages for whom English is the communicative medium of choice, and often the only option" (Seidlhofer, 2011:7, emphasis removed). This conceptualization is functional rather than formal: the norms of interaction are established in interaction and regulated by the speakers themselves rather than following external (NS) principles. ELF users are commonly found to strive toward achieving mutual intelligibility and getting things done, and in doing so, they accommodate and adapt to their interlocutors and the context, monitor comprehension and collaborate to achieve understanding (e.g., Seidlhofer, 2011;Mauranen, 2012).
When describing speakers who use ELF, the research refers interchangeably to ELF speakers or ELF users, albeit recently "ELF speaker" has grown unpopular because it may suggest a view of ELF as a variety of English. By "user", researchers refer to a person who is "capable of speaking the language and successfully navigating through different kinds of interactions in a diversity of communicative environments" (Llurda, 2017:520e521). A competent user has experience of this kind of use, and s/he has therefore developed an awareness of interacting in various situations (ibid). For ELF users, this means particularly that they are to be considered competent users of English in their own right.
This article explores speakers' emic orientations to language in domestic interactions between participants who could be described as ELF users. It seeks to explore emically relevant categorizations visible through sequence and repair organization and orientations to language expertise, and aims to investigate which speaker labels would best suit the interactions. Before moving on to the analysis, it is important to review relevant research from CA-SLA and ELF to understand what kinds of observations researchers have used to justify their labelling practices.

Relevant research from L2/NNS interactions
Language learning is a recognizable activity for participants (Pallotti and Wagner, 2011), but L2 speakers/NNSs are not doing being language learners at all times (e.g. Firth, 2009); they do not necessarily orient to learning as an activity or demonstrate their participation in the interaction as learners. The conversation analytic approach can explore learning from two perspectives: by observing changes in participants' practices over time, or by detecting instances where participants orient to gaining control of linguistic resources (Pallotti and Wagner, 2011:4). The majority of CA-SLA studies use the latter approach in institutional contexts, even though language learning also occurs outside of classrooms. In this section, I review some CA-SLA research on L2/NNS interactions outside the language classroom, scrutinize the types of activities that have been interpreted as conveying an orientation to NNS identity in these contexts, contrast them to ELF research findings, and look at the authors' naming practices for their participants.
By investigating Finnish NS-NNS interactions between secretaries and students of a language school and between friends, Kurhila (2005) found that while NNSs oriented to linguistic forms as repairables, NSs treated these instances as either NNSs' self-repair or as self-initiated repair initiations that only prompted quick repairs from the NSs. The NSs also ignored NNSs' word search attempts: they indicated understanding with acknowledgement tokens/nods which attempted to close the side sequences that oriented to language. Similar findings were observed in Theodorsd ottír (2011), where L1 speakers of Icelandic indicated understanding at an early stage, while their L2 interlocutors insisted on bringing their turn constructional units into completion. A slightly differing tendency was found in Wong (2004), where L1 speakers of English were observed to allow more "room" (p. 129) for L2 speakers to finish their turns by delaying their own responses. Kurhila (2005) contends that NSs are reluctant to portray themselves as language experts because it would make the linguistic asymmetry between the speakers salient. She further suggests that the context influences how the asymmetry is treated: in her data, the secretaries' main objective was to achieve the institutional goal, not to act as language instructors. In interactions between friends, Kurhila argues, NSs attempted to maintain their role as a friend and "avoid 'being second language teachers'" (2005:156). Similar observations were also made in Wong (2005) which investigated three NS-NNS interactions in English. There were no linguistic asymmetries made visible "since NS appear[ed] to sidestep or disattend to NNS's ungrammatical troublesource" (p. 160). In this way, Wong argued, the NSs avoided exhibiting their linguistic expertise. The NSs also opted for repair rather than other correction in response to the NNSs' troublesome utterances, thus assuming responsibility of communicative turbulence. Wong contrasted these findings to classroom interaction, where "the teacher and the [NNSs] jointly orient to and build the talk in a different manner, constructing their identities as linguistic expert and novice, native and nonnative speakers" (p. 160). The paper shows that in a classroom context, interaction is differently organized than in naturally occurring NS-NNS interactions e it focuses on form and accuracy rather than on meaning. In Kurhila's, Theodorsd ottír's and Wong's papers, then, the NSs were doing not being language teachers e their focus was on the interactional progress e whereas the NNSs in Kurhila's and Theodorsd ottír's papers showed a form-focused orientation to their own non-nativeness.
In another paper (2004), Kurhila examined L2 speakers' interactions with L1 speaker office/information/reception desk workers. The interactions were institutional by nature: the clients arrived at the desk with a task (e.g. information requests, filling a form) and left after it was fulfilled. In these interactions, too, the L1 speakers systematically adhered to their institutional identities, while the L2 speakers (the clients) oriented to their (lack of) linguistic skills. In this paper, Kurhila identified the following practices which, she argued, displayed the L2 speakers' status as language learners: Brouwer (2003), however, argues that word searches alone do not necessarily indicate an orientation to language acquisition, and for this to be the case, two conditions should be fulfilled: "(a) the other participant is invited to participate in the search, and (b) the interactants demonstrate an orientation to language expertise, with one participant being a novice and the other being an expert" (2003:542). In Kurhila's papers, then, it was often the L2 speaker who displayed such an orientation, while the L1 speaker avoided such indications.
Comparing these findings to ELF research, we see that although the practices identified by Kurhila (2004) may in some ways signal NNS identity, they may not always be linked to language learning as an activity. Collaborative word searches, verbalizations of lack of knowledge and requests for help are by no means uncommon in ELF (Mauranen, 2012), and they are particularly found in connection with codeswitches to other languages that typically request for lexical help from other participants of the same linguistic background (Hynninen et al., 2017). Although speakers often seek for linguistic help from each other, the help is routinely performed in a collaborative manner without a clear role divide between linguistic experts and novices (Mauranen, 2012). This observation, together with the finding that linguistic other-corrections are rare in (academic) ELF interaction (Mauranen, 2012:217), suggests that Brouwer's second condition is not unequivocally fulfilled in ELF interaction. This may indicate that the kind of expert-novice discrepancy so often observed in NNS-NS interactions may not be so relevant in interactions where the majority of the speakers are NNSs. The two aforementioned ELF studies used a corpus of academic (not language classroom) interaction, comprising primarily NNS-NNS interactions where the speakers' English proficiency was relatively high and the interactions were particularly content-related (Mauranen, 2012).
Another point of comparison between NNS-NS and ELF interactions is the use of explicitness strategies. In a dinner table discussion between L1 and L2 speakers of English, Gardner (2004) found that there seemed to be a slight mismatch in timing of questions and answers, and although the L2 speakers displayed no difficulty of understanding the L1 speakers' questions, they took a longer time to answer. Some of the L1 speakers in the study extended their question turns with post-expansions that sharpened or refocused the question, thereby adding greater explicitness to their questions to the L2 speakers. This kind of enhanced explicitness is also a common observation in ELF interaction (sometimes referred to as accommodation, see e.g. Seidlhofer, 2011). Explicitation strategies such as paraphrasing, topic negotiation, repetition, and left and right dislocation (placing a focal element outside the clause structure; see e.g. Mauranen, 2012) are considered common features of ELF communication (see Pietik€ ainen, 2018a for comparable strategies in ELF couple interaction). Explicitation used by the L1 speakers in Gardner's data is therefore not specific to L1 speakers communicating with L2 speakers but could be viewed from a larger perspective as a communication strategy for speakers in intercultural situations where there is a risk of miscommunication due to variable linguistic backgrounds.
While ELF users are found to sometimes ask for linguistic help from their peers and thus reveal their NNS identity, commonly ELF communication has been found to proceed in a mutually helpful manner, employing explicitation practices similarly to L1 speakers in Gardner's paper. This leads one to ask whether a clear-cut division to L1 and L2 speakers is so well justified after all. In all the CA-SLA papers referred to here, authors referred to the speakers as L1/L2 4 or NNS/NS. In those papers where speakers displayed hesitation of linguistic forms and overtly oriented to the linguistic asymmetry between themselves, the NS/NNS division seems to be supported emically, while in other papers (e.g. Wong, 2005) this distinction was not clear.
Reflecting against this background, the current paper analyses an informal context where speakers are socially in principle equal while from the researcher's perspective, they can be described as using English as their common lingua franca. In order to explore the extent to which language is oriented to as a "learnable" in the data, the paper first seeks to answer: How common is sequential orientation to language (e.g., grammar, lexis, pronunciation) compared to overall sequential orientation in the data?
Second, the paper focuses on repair organization and asks: 1.
Verbalizations of lack of knowledge ('I don't know how to say in Finnish') 2.
Requests for help ('How do you say', 'What is it [called]' etc.) 3.
Seeking for collaborative word search by repeating syllables while shifting gaze to the L1 speaker (Kurhila, 2004:59) How is repair organized in the interactions and to what extent are linguistic matters repaired? Are verbalizations of lack of knowledge, word-search, and requests for linguistic help as characteristic of the data as they are in Kurhila (2004)? Do the participants demonstrate such an orientation to language expertise that one is a novice and the other an expert? (following Brouwer, 2003:542).

Data
The materials investigated come from a collection of seven interlinguistic couples' audio recordings collected for a longitudinal research project in 2012e2018. They consist of 66 h 35min of self-recorded conversations from couples living in Finland, Norway, the UK, Canada and Zambia. 5 All participants had reported some other language than English as their L1. For this paper, I analyzed one randomly chosen recording from each couple from the latest (2018) set of recordings, creating a subset of approximately 3 h of talk. The data was transcribed using slightly adapted Jeffersonian conventions (see Appendix A). Only recordings that involved longer stretches of talk between the couples without their children were selected, which is why the subset only involves recordings from six couples. The interactions were mostly carried out in English which was the only or most fluent language the couples shared in the beginning of their relationship, although some language alternation and turns/sequences carried out in other languages are observed.

Method
To explore the first research question, the overall sequential orientations were traced. That is, using conversation analytic means, activities performed or topics discussed in each sequence were identified. A sequence is here defined, following Schegloff (2007:xi), as a stretch of talk that hangs together; a (possibly) multi-turn unit arrangement that runs from sequence (or topic) opening to its closing (or topic transition). Sequences can involve insert expansions, side sequences, and postexpansions e these were noted separately when they involved repair work or if the participants oriented to linguistic aspects.
Then, a turn-by-turn analysis of repair initiations (incl. word search) was performed. Self-repairs i) immediately following the troublesource, ii) at the transition relevant place after the trouble-source turn, and iii) in third position or later were identified, while other-initiated repairs were classified to i) those initiated with a question token and ii) others. Self-initiated other-repairs were relatively uncommon and were noted separately. Then, a standard/deviant case analysis of repair organization was performed, while repair related to linguistic form was analyzed separately. To explore the possible novice/expert orientation, attention was paid to who initiated and performed linguistic repair and in what ways.

Sequential orientation
The analysis of sequential orientations found very little orientation to language overall. Typically, in each interaction, there was an activity 6 that functioned as a frame or grid around which (story)telling activities, discussions, summonseanswers, and other units of social action were organized. In the data, such overarching activities were: driving the car/giving directions (Laura & Thomas), watching a TV program and commenting on it (Sanna & George), having breakfast with the family (Chun & Nils), reportings of the day's doings (Minna & Henrik), making preparations for a dinner party (Carmen & Kjetil), and baking an Easter bread (Elisa & Budi). 7 For space limitations, only the last couple's sequential organization will be used as an illustration of the orientation to an overall activity.
Appendix B describes the main orientations in the interaction of Elisa (E) and Budi (B) at each moment of recording (rounded off to the nearest 5 s). The interaction was recorded in the evening before Easter Sunday. It is readily observable in the data how the action of baking Easter bread is oriented to as the underlying course of action, while other topics are oriented to in-between. The recording starts with Elisa opening ingredients packages while Budi asks how long it will take (to bake the bread). Elisa suggests that he might help her, and the couple's interaction revolves around the preparation and ingredients of the bread. The baking as a verbally oriented-to topic is set aside at 2.20 when Budi remarks that the children liked the Easter egg hunt. This topic is then taken up and the discussion turns away from the baking activity, although Elisa can be heard preparing the dough. The baking activity is again oriented to at 10.00, where Elisa starts shaping the breads. After about a minute, the talk turns back to the children and other tellings. At 21.30, Elisa initiates a new sequence with "uhh, so I'm thinking to boil eggs, as well, that go in the middle of the bread", thereby orienting to the topic of Easter bread but at the same time initiating a new task. This suggestion is contested by Budi, who suggests she should do it in the morning instead. A lengthy discussion of dyeing eggs ensues, and then other Easter preparations are discussed and performed. The couple 5 For more information on the couples' backgrounds, see Pietik€ ainen (2017, 2018a). 6 In defining 'activity' I follow Heritage and Sorjonen's (1994:4) definition of it as "the work that is achieved across a sequence or series of sequences as a unit or course of action" which is relatively goal-oriented and topically coherent, such as "talking about the weather". It can be moved into and out of recurrently in the course of interaction. 7 All names are pseudonyms.
resumes the topic of baking for the last time at 28.20, when she enquires his opinion on whether the breads will rise too much if left overnight, and whether she should put eggs in the middle to make the shape. He rejects these suggestions and moves swiftly to complementing her for the preparations, which seems to function as a pre-closing device for the whole discussion.
Remarkably, neither the overarching activities nor the overwhelming majority of the shorter sequence units in-between focused on the speakers' linguistic production. This tendency was present in each couple's data, while it would likely have been reversed were the data gathered from a language classroom. Only on very few occasions was language topicalized over several turns. Extracts (1) and (2) present two such deviant cases.
In (1), the watching of TV is oriented to as a social action by frequent commentings of the TV show. Sanna's first turn takes place after a TV host's extended introduction. It enquires what language the TV host is using, thereby assigning expertise on Zambian languages to George. George's answer (line 03) is delayed but matter-of-fact, the high pitch and lack of hedging suggests that he assumes the answer to be self-evident; he is thereby exhibiting his linguistic expertise over English. Sanna disagrees laughingly (line 05), contesting George's knowledgeable position, after which George utters a quiet "mhm" (line 07). Following a rather long pause during which someone can be heard walking on heels on the TV, Sanna treats the sequence as closed as she begins a new evaluative turn. George, however, overlaps (line 10), returning to the previous topic by assessing the language to be Xhosa or Zulu (immediately repaired to include a click sound in Xhosa, marked with a €). After a brief pause (line 11), he arrives at a hedged conclusion, "i think it's zulu.", displaying tentative knowledge over African languages and simultaneously, orienting to Sanna's earlier turn (line 05) as accurate. It becomes apparent that both interlocutors have knowledge over English, but clear dominance is not assumed by either party, while George is assumed to have more knowledge over African languages. It is also interesting that albeit the sequence shows clear negotiation of language expertise, the language evaluated is not that of either of the interactants but that of the TV host. In fact, there are several occasions in the recording where Sanna comments on or imitates linguistic matters such as pronunciation or word choice of the TV characters, but the couple refrains from discussing linguistic matters related to their own talk.
(1) Sanna and George are watching a Zambian dating show. (2) Chun and Nils talk about a fabric patterning Chun did on her son's shirt. In (2), line 01 Chun answers Nils's question "what did you do:" (posed some turns earlier) with a hedged "broch e:,". There is a long pause during which Nils can be heard to chew food, before he asks for a clarification: "brosh?". He then modifies this clarification question to concern the spelling of the word, which is overlapped by Chun's repetition of the troublesource. In lines 05e06 he self-repairs the other-initiated repair to concern the pronunciation. Immediately after this turn, he is heard to practice the pronunciation (line 07). In line 08, there is a micropause which opens the possibility for turn-transition, but as Nils continues to practice the pronunciation of the word by returning to a form which is closer to his original enunciation (see line 03), Chun's approval of the pronunciation he produced in line 07 is overlapped with this new turn (lines 09e10). Again, there is a short pause after which he again produces the pronunciation modelled by Chun, while Chun overlaps with a repair of the pronunciation he produced in line 09. It is evident that in this part of the sequence, Chun is oriented to and assumes herself to be the authority of linguistic knowledge concerning the handicrafts on the shirt. However, in the next part, the roles are reversed. After a 2.4 s pause which could mark the end of the sequence, Chun refuses the choice of word and engages in a new word search (line 15). The word search remains solitary even during a rather long pause (line 16), until Chun produces a new candidate, "BRO:dery.". Nils latches to this turn with a change-of-state token and a candidate understanding in Norwegian: "bro:-(.) bro:deri:" (line 18). He quickly adds another change-of-state token and affirmations, which Chun's confirmation overlaps. In line 23 he repeats the Norwegian word with a deep voice which seems to affirm his epistemic status. In lines 25e26, with a few false starts, he comes up with an English translation for the word, which Chun practices (line 27) and he repeats (line 28). However, it seems that Chun is not entirely convinced of this word choice, as she tries another option "a crochet" in line 30, which she quickly abandons, returning to sampling the word (em)broidery (line 35). The role-reverses of language expertise in this extract demonstrate how flexibly the knowledgeable status can be assigned and adopted in the couple interactions. However, this kind of multi-turn collaborative word-search is very infrequent in the data and is only observed once more in extract (12).
Looking at the sequential orientations in the data, the first research question can be answered as follows: sequential orientation related to language is uncommon. Furthermore, orientation to language expertise in the few sequences where language matters were topicalized shows no sign of clear expert/novice distinction (except what will be seen in (12)). The next section focuses on the repair organization in the data to explore the extent to which linguistic issues are repaired and whether participants show novice/expert orientations.

Repair organization
Contrary to Schegloff et al. (1977) that finds other-initiated repair highly constrained, in the current data, repair initiations by self are more common than repair initiations by other, but the difference is not drastic overall. In fact, in one couple's data (S&G), repair initiations by other are more common than repair initiations by self. The next two sections analyze self-initiated repairs and other-initiated repairs respectively.

Self-repair and self-initiated repair
The object of repair was uncommonly language and hardly ever aspects of the English language; in self-repairing, the speakers more commonly addressed possible troublesources related to intersubjectivity, e.g., misspeakings (3), inaccuracy (4), and misinterpreted implications of their earlier turn(s) (5). Only rarely were such linguistic objects self-repaired that could point toward the correction of a linguistic error in English vocabulary (6), morphology/syntax (7), or pronunciation (8).
The infrequent self-corrections concerning language often corrected other languages than English. In (9), Kjetil repairs his turn in Spanish (his wife's L1), whereas in (10), and (11), speakers restart turns in their own L1s.
As can be seen here, self-corrections are observable both in the less fluent languages of the speaker (as judged by speech perturbations), as well as so-called L1s. According to Schegloff et al. (1977:370), such word replacements are not uncommon in native English data either. There is not enough evidence to suggest that the self-corrections in the data, e.g., (3)e(5),(9), result from an orientation to nonnative speaker identity (but see (12)). Rather, they result from the speaker identifying a problem in something they have uttered previously when contrasted to what they project to say next. Restarts and hesitations are a common feature of unplanned spoken interaction and can therefore not be straightforwardly taken to imply NNS orientation.
Furthermore, there were no overt verbalizations of lack of linguistic knowledge or language-related requests for help concerning English in the way suggested by Kurhila (2004), nor did the participants display novice/expert orientations (Brouwer, 2003), except in a fluctuating manner in (1) and (2). Only one other sequence displays such an orientation regarding another language than English. In (12), Carmen displays her identity as a language learner of Norwegian by both inviting Kjetil to participate in the word search and orienting to him as the expert (cf. Brouwer, 2003). In this extract, Carmen is commenting on a Norwegian cookbook of Mexican food. In lines 01e02, she initiates a telling of churros sold on the streets of Mexico being much longer than those in the book. The pause before "på vei?" and the interrogative intonation (line 02) display hesitancy of this formulation, and albeit Kjetil indicates understanding (line 03), Carmen initiates self-initiated other-repair in line 04 by altering the pronunciation of the troublesource and adding an interrogative "eller?". It is here clearly observable that Carmen treats Kjetil as a language authority of Norwegian, and Kjetil exhibits this epistemic status by producing a repair (line 04). However, in the next line, it becomes apparent that "på veien" may not be the right candidate for the word-search, as Carmen translates the looked-for item into English (line 07). Kjetil provides a new repair which he hurriedly restarts (line 08), perhaps as he realizes what Carmen really intends to say, namely "churros (sold) on the street" rather than "churros on the way". Carmen accepts this repair (line 09) and continues the telling by referring to the book and assessing the churros to be "veldig litten", very small, again ending with an interrogative intonation. After a short pause, she even urges Kjetil to look at the picture, which he apparently does, accepting her assessment (line 12). In the next line, Carmen contrasts the picture in the book to her experience of Mexican churros (notably, self-correcting to follow the linguistic advice received from Kjetil, see line 13) and even before she gets to the end of the utterance, Kjetil latches with a suggestion for a description of the Mexican churros, pronouncing both syllables of "lange" clearly and with emphasis (line 14). This suggestion is emphatically endorsed by Carmen, and even in the last line, she repeats this word several times, which is a way of adding emphasis characteristic to Carmen (Pietik€ ainen, 2017:141).
Most interesting in this extract and descriptive of the overall data set is that participants avoid other-correction of linguistic form. Even in this extract, Kjetil does not address the non-standard features which Carmen does not request him to correct (e.g., lack of reversed word order in lines 01e02 and 09, mispronunciation of her ("ha:r") and liten ("litten"), use of the word liten instead of små or korte, and lack of plural markers in several other places). Language as the object of learning activity does not become salient unless the "learner" herself initiates such an activity by overtly initiating other-repair. As this type of interaction is only observable in this specific sequence, the speakers in the data set cannot be generalized to identify as language learners on the whole, and especially not regarding English.
When it comes to Kurhila's (2004) third aspect of nonnative talk, collaborative word searches by repeating syllables with a gaze fixed to the interlocutor, the current audio data does not allow information on gaze. However, word-searches where syllables are repeated, were uncommon. In word search, words preceding the searched-for word were repeated in their entirety (13), or the utterance was reformulated with a word insertion (14). In both types, the current speaker finishes the utterance even if it takes a longer time, without the other party offering collaborative completion.
Related to word-search, the speakers' turns may often be described as verbose. In this type of social talk, getting fast to the point seems to have less social value than engaging in conversation and achieving mutual intelligibility and even consensus. 8 The speakers often rephrased and reformulated their utterances and added explanations and examples, see (15), (16). In addition, recycling of crucial words was common, such as the recycling of "upstairs" in (17). These practices can be interpreted as the kind of explicitness used by L1 speakers in Gardner (2004) or by ELF users in ELF literature (e.g., Mauranen, 2012.

Other-initiated repair
Other initiated repair commonly with an open class initiation (e.g., hm, huh, uh, what, sorry, h€ ah, mit€ a, mik€ a) which usually seemed to repair trouble with hearing or paying attention, see (18), (19). Category-specific interrogatives (e.g., who, which one, why), see (20), and otherwise specified clarification requests, see (21), (22), seemed to relate directly to achieving intersubjectivity rather than to language issues. Others also corrected factual matters, see (23), (24). Language-related other-initiated repairs were uncommon, which reflects the interlocutors' reluctance to display language expertise over other, similarly to Kurhila's (2004Kurhila's ( , 2005 and Wong's (2005) NS participants. Also, embedded linguistic repairs of "non-native-like constructions" (see Brouwer et al., 2004:78) were not observed. 9 Rather than producing the correct form in the next turn, the second speaker routinely focused on providing a sequentially relevant next turn. For example, in (25) Henrik does not answer Minna's question by correcting the right syntactic form (e.g., "it'll be ready in November"), instead, he just provides the factual answer: All in all, the analysis of the couples' repair organization shows that self-repair more commonly addresses troublesources related to intersubjectivity than linguistic issues, and when linguistic matters are self-corrected, these often relate to other languages, also the speaker's own L1. No overt verbalizations of lack of linguistic knowledge or requests for help were observed, except in one word-search sequence concerning Norwegian (12). Even in this sequence, the expert refrained from other-correcting the novice on linguistic matters which the novice did not mark as correctables. Word searches were routinely solitary in nature, and the other party commonly allowed the speaker to reformulate and rephrase their utterances. In this way, the participants both allowed room for each other to finish their sentences and used explicitness practices, like the L1 speakers in Wong (2004) and Gardner (2004) e and as commonly found in ELF.
Other-repair was common, and it routinely addressed intersubjectivity-related matters such as trouble with hearing/ attention or accurate understanding. Although factual mistakes were corrected, the speakers avoided linguistic othercorrection (even embedded correction), and rather focused on offering the sequentially relevant next turn. This shows an orientation where the speaker identities are perceived more or less equal and rather those of a legitimate (multilingual) user than a learner (of English).

Concluding discussion
From an etic perspective of the researcher, the data examined here displays many features that are commonly attached to NN/L2 talk e there are syntactic and clause constructions that do not adhere to standard English, hesitations, restarts, use of other languages etc. However, when investigated from an emic perspective, we see that the interlocutors' rights to speakership are unproblematized. Overall, sequential orientation to linguistic issues is very rare, and most times conversation flows around a shared activity which frames shorter sequences of tellings, plannings, opinion exchanges etc. On the few occasions when matters concerning the English language arise (see (1) and (2)), ownership over English alternates and both parties assume a degree of knowledgeability.
Although self-repairs are more common than other-repairs in the data, the difference is not very drastic in the couple interactions, in contrast to Schegloff et al. (1977). The troublesources most commonly repaired by self concern misspeakings (3), inaccuracy (4), and earlier turns misinterpreted by the other party (5). Other-corrections most commonly relate to achieving intersubjectivity or repair of factual mistakes, while language-related other-initiated repairs are rare. Rather than engaging in linguistic hairsplitting, the participants thus routinely provide a sequentially relevant next turn. Self-repairs that may have to do with language (6e11) are also uncommon overall and do not concern only English; such repairs also occur in so-called L1s and other languages. Self-corrections may therefore be seen as related to general utterance formation and should not be automatically considered an indication of NN/L2 speakership.
Verbalizations of lack of linguistic knowledge, collaborative word-searches and overt requests for help on linguistic matters were seldom, in contrast to Kurhila (2004). Only one sequence (12) contains word-searches where the other party is invited to correct the speaker's linguistic production and is clearly oriented to as the expert, as in Brouwer (2003). Yet, the linguistic expert retains from correcting nonstandard features which the novice does not explicitly mark as troublesources. Furthermore, the troublesource language is Norwegian e the L1 of the "expert" e not English. As in the interactions between friends in Kurhila (2005), the couples avoid being language tutors of the shared lingua franca. This may be due to other identities (spouse, parent) being more relevant in their interactions and perhaps even that an expert/novice divide would create an unnecessary shift in power dynamics in a language which they otherwise regard as equal (Pietik€ ainen, 2014).
As regards to the emic orientation toward English especially, in the majority of the interactions, it is the main language of communication and not oriented to as a foreign language. Nonnativeness or nativeness of English is not an issue that in any way seems to have importance to the interlocutors in the immediate context, and hence such categories as NS or NNS are made redundant. However, other languages and language alternation are also a part of the naturally progressing interaction. The more recent conceptualizations of ELF which assume equal rights to speakership and include translanguaging and the possibility of using other languages than English (e.g., Jenkins, 2015) seem to fit the interactions explored here. It is also made obvious that the speakers have individual and shared multilingual repertoires which they put to use in various degrees. Therefore, the interactions can equally well be described as trans-or multilingual. Numbering the languages in the interactions would make little sense and if anything, English would be the number one language of communication between these pairs. The label L2 would hence reduce the interactions to something they are not.
Apart from the linguistic perspective, the data evidences the speakers' close relationship to each other: despite the obvious parent talk (e.g., speaking with/about their children), their interactions display a shared social category (e.g., the use of the pronoun "we" and references to a shared history). It is therefore emically justified to refer to the interactions by the category of "couple" or "family". Another obvious way of labelling the data emically would be to refer to the sequential activities performed overall or in shorter sequences; see section 5.1.
It is evident that refraining from speaker labelling until a participant orientation analysis has been performed allows for a more neutral standpoint for objective investigation as well as relies more substantially on the emic principles of CA. In examining interlinguistic interactions, treating them primarily as lingua franca interactions until otherwise shown may also provide a more objective outlook for the analysis. While the temptation prevails to use such labels as L2 speaker and NNS to describe participants whom we factually know to have a different L1, the participants may not actually do being NNS in the interactions, and therefore it would be ethically unjustifiable to take the easy way in speaker labelling. If participants are assumed to be NN/L2 speakers from the outset, a risk prevails that the analysis may then focus too heavily on features that have little emic relevance to the participants (e.g., speech perturbations). Another risk with such an approach is that observable orientations to equal linguistic expertise may be deemed superficial (cf. Wong, 2005) when they contrast the researcher's a priori assumptions. A change to a more emically justified labelling perspective may require analysts to alter their data management practices significantly. A better alternative may be to collect a wider range of interactions, perform an emic analysis of participant orientations within and across the data and only thereafter focus on a selected data set where topics of interest are de facto oriented to by the participants. This kind of a practice has of course been used in classical style CA research, but less so in CA-SLA and ELF.

Funding
This work was supported by Alfred Kordelin Foundation and The Finnish Cultural Foundation (Grant: 00200087).

Conflict of interest
There is no conflict of interest. syllables not recovered ((mobile)) author's comment Italic ((Spa)) pragmatic translation into English from mentioned language (e.g., Spanish) Last tasks given for the evening, setting the table. E assumes the role of the organizer ("but first, table needs to be cleaned", "so next is…", "and then…"). B reports on progression and provides positive feedback on several occasions "they're gonna enjoy themselves". 26.55e28.20

Appendix B
Admiring task cards E has prepared for the children. 28.20e29.50 Final details on the bread, positive evaluations of the preparations. Closing with E: "okay, let's go to bed" B: "okay yeah, time to go to bed." E: "good night" 29.50e30.00 Orienting to the tape recorder, waiting until it is exactly 30.00, then stopping it.