Using stimulated recall to explore the use of communication strategies in English lingua franca interactions

1.1 Approaches to analysing communication strategies in ELF interaction

Researchers who study communication strategies in ELF interactions have employed a number of approaches to and models of interaction analysis. A number of studies on communication strategies (CSs) in ELF interactions explicitly or implicitly draw on Conversation Analysis to describe and explain the use of CSs (e. g. Firth 1996; Kaur 2010). An essential principle of conversation analytic methodology is social interaction as a joint accomplishment, where “action and meaning are co-constructed” by the interlocutors (Jenks 2012: 391). In Conversation Analysis, findings about particular social interactions are derived from the data, which consist of transcripts of interactions, analysed in detail. Moreover, little to no background information about the participants, goals or larger context of the interaction is considered, and no predetermined categories are used. Therefore, CSs are not separately defined or classified. Researchers identify practices and patterns which occur, analysing what is accomplished interactionally through the use of these patterns and practices. For example, Schegloff grouped all “practices for dealing with problems or troubles in speaking, hearing, and understanding the talk” as accomplishment of repair (Schegloff 2000: 207).

Among other analytical approaches used to explore CSs in ELF interactions are accommodation theory (Cogo and Dewey, 2006), interactional discourse analysis, and analysis of ELF users’ pragmatic routines (e. g. Björkman 2011; House 2003). These approaches can incorporate background information about interlocutors, goals, context and other situational features of the interaction. Indeed, researchers have noted differences in CSs observed across different contexts. For example, Mauranen’s (2006) data from academic settings show that difficulties in understanding are almost never ignored (i. e. let it pass strategy), as they sometimes were in Firth’s (1996) data from business settings.

1.2 Investigating ELF use and context

Mortensen (2013: 38) suggests that the use of ELF is “one out of several different components that constitute a speech event.” Therefore, other contextual factors should also be taken into account when explaining how ELF is used. This has been done in multiple ELF studies, including research analysing CSs. For example, Guido (2012) analysed ELF interactions between African asylum seekers and Italian intercultural mediators while incorporating factors such as interlocutors’ typologically different first languages (L1s) and different cultural schemata and the marked power imbalance between interlocutors. In another study, Jenks (2012) linked the use of reproof or “reprehension” strategies to the particular “norms, expectations, and interactional and institutional goals of the communicative context” (Jenks 2012: 402), that is, voice-based chat rooms. In these studies, the communicative behaviour is analysed as interaction between ELF users which is shaped by, for example, the particular setting, interlocutors’ characteristics, and communicative goals.

1.3 Communication strategies to prevent or repair difficulties in understanding in ELF

Among many possible CSs in ELF interaction, the current study targets solely those used to prevent or repair difficulties in understanding. Although CSs can potentially be defined in many ways, most research cited below has used the interactional approach to identifying CSs (e. g. Firth 1996), namely, CSs occur when interlocutors strategically draw on their available resources to jointly negotiate and implicitly agree upon meaning. In this section, I present previous research on the use of these strategies to prevent or repair difficulties in understanding in ELF interaction.

Early research occurred in the context of the workplace. Firth (1996) analysed English phone conversations between European, Asian and Middle Eastern businesspeople. He found that when part of one interlocutor’s utterance seemed not to be understood by another interlocutor but was not crucial to the overall message, the second interlocutor would frequently “let it pass” unless it later became essential to understanding the message. Pitzl (2005) examined business meetings between European and Asian ELF users, targeting instances of non-understanding, “when the listener realises s/he cannot make sense of an utterance” (Bremer 1996, as cited in Pitzl 2005: 52–53). In these meetings, the few obvious cases of non-understanding were not “let pass” but were typically signalled by repeating problematic words or phrases. The meaning of the words or phrases was then negotiated by rephrasing or elaborating on the words or phrases in question.

Much recent research on difficulties in understanding has taken place in academic settings. For instance, teachers of English from different Asian countries taking a course in Singapore were recorded chatting to each other (Deterding and Kirkpatrick 2006; Kirkpatrick 2007). Few difficulties in understanding were evident, but those which did occur were usually signalled by clarification or repetition requests (Pardon?) or inappropriate responses to questions. The interlocutors then negotiated the meaning of the initial utterance through rephrasing or elaboration. When interlocutors did not explicitly signal lack of understanding, remaining silent or providing neutral backchannelling (Mmm) after hearing the trigger utterance, difficulties in understanding tended to remain unresolved. Interlocutors also used strategies to prevent difficulties in understanding, such as naming topics or topic changes, thus making them explicit in the interactions.

Kaur (2010, Kaur 2011, Kaur 2012) investigated how international graduate students at an English-medium program in Malaysia worked to pre-empt or repair difficulties in understanding during naturally occurring conversations and discussions. Interlocutors repeated or rephrased their own utterances following both implicit and explicit signs of difficulty in understanding (e. g. prolonged silence or clarification questions, respectively). These strategies led to apparent resolution of the difficulty in understanding (see also Matsumoto 2011).

Several studies have documented both types and distributions of particular strategies used by ELF users during post-secondary classes, lectures, seminars and group-work sessions (Björkman 2011; Smit 2010), particularly when faced with difficulties in understanding. For example, Mauranen (2006) found that difficulties in understanding were explicitly signalled through clarification questions or repetitions of problematic words, but most cases of misunderstanding were not indicated in targeted or focussed ways. Interlocutors responded to these difficulties in various ways. Synonyms were provided, statements were rephrased or sometimes elaborated on. ELF users would also work to prevent potential misunderstanding in various ways, by checking interlocutors’ or their own comprehension (e. g. Yeah? or Did I understand right?) or very frequently by self-repairing their own utterances (e. g. It will be no questions eh it will be no conflicts). In a longitudinal study, Smit (2010) explored the occurrence of and reaction to repairables, that is, “what is being repaired” (Smit 2010: 168) in classroom interactions. Repairables were not identified in isolation, but only within interactions as the target of repair in the course of “negotiating or constructing understanding” (Smit 2010: 169). At the beginning of the program, students and especially teachers did target mishearings (unintelligible speech) for repair by using clarification requests. However, as the program continued, repair related to mishearings greatly decreased, which Smit attributed to teachers and learners becoming more familiar with the academic content and with each other’s speech. Like Mauranen, Smit noted that the ELF users consistently demonstrated a commitment to understanding and being understood. When difficulties arose in understanding a message or topic, ELF users typically addressed the difficulties at the moment they appeared.

Findings on CS use for difficulties in understanding in ELF show some similarities as well as some differences. In many cases, ELF users repeated, rephrased or elaborated on utterances when interlocutors signalled difficulties in understanding, although different combinations of CSs were used in different studies. In most studies, ELF users worked to prevent, not simply address, difficulties in understanding. However, in some studies, ELF users addressed every difficulty, while in other studies, occasionally users did not explicitly signal that they had difficulty understanding other interlocutors. Much has been learned about the frequency and kinds of CSs used in particular contexts for ELF interaction. However, the relative use of specific CSs, such as let it pass, may not be as significant as the way that particular CSs are used in a given interaction. One aspect of the interaction which can shape the use of CSs by ELF interlocutors is the communicative context.

1.4 Links between context and communicative behaviour

Reviews of research in ELF contexts (e. g. Jenkins et al. 2011; Murray 2012) tend to emphasize the commonalities in CS use. However, as Mortensen (2013) notes, other contextual factors also contribute to communicative behaviour. For instance, Kirkpatrick (2007) suggests that a cultural notion of “face” shared by the ELF users in his study may have shaped their CS use. Nevertheless, the role of interlocutors as a contextual factor has rarely been investigated in ELF interactions. One way to access interlocutors’ knowledge is to examine interlocutors’ memories of their cognitions (thoughts and feelings) during interaction. Methodologies which elicit these types of memories rely on stimulated recall as a research tool (Mackey and Gass 2005). House (2003) incorporated stimulated recall to explore CS use in ELF interaction. She audio-recorded international students in Germany, who engaged in discussions based on a researcher-provided prompt. Several weeks later, House interviewed each student, who listened to excerpts and read transcripts of instances indicating “imminent interactional trouble” (House 2003: 567). In the stimulated recall, students had different explanations for one commonly observed strategy, namely, repetition of the last phrase or word from their interlocutor’s previous turn. One student suggested that repetition helped her remember and understand the previous speech and show that she was listening attentively. For another student, repetition helped him respond to the previous turn.

It is clear that, moment by moment, repetition across turns can be used for different communicative functions in ELF interaction. Using detailed interaction analyses, Cogo and Dewey (2006) and Cogo (2009) also found that this type of repetition was used to support other interlocutors. However, by eliciting interlocutors’ memories of the interactions, House (2003) was able to identify multiple, interlocutor-specific purposes for using that strategy, some of which were related to understanding and some of which had to do with politeness or claiming a turn. Eliciting the interlocutors’ cognitions thus allowed House to integrate her interpretation with interlocutors’ own accounts of how repetition was used, resulting in a more comprehensive analysis. These findings suggest that ELF users have different reasons for using similar CSs, even in similar contexts; therefore, the communicative effect of CSs should ideally be determined not simply by their frequency, distribution or context, but also by interlocutors’ interpretations of how CSs were used in that particular interaction.

1.5 ELF use and naturalistic interaction

In most published studies on ELF interaction, data are taken from conversations or interactions which occur naturally in those contexts, rather than being arranged by the researcher. Seidlhofer’s (2011: 7) definition of ELF suggests that when interlocutors of different L1s choose to use English to communicate with each other, English is being used as a lingua franca. The research site for the current study was an English-medium university, where English is the default language for interacting with (an) unfamiliar interlocutor(s), the unspoken assumption being that English, as opposed to another language, is the language in which all potential interlocutors can functionally communicate. It is in this light that the unscripted interactions in English described in the current study, elicited at the English-medium university from researcher-designed communicative tasks, are presented as authentic use of ELF in interaction, though the interactions were not naturalistic.

Questions have been posed about the general applicability of findings from ELF interactions in communication which did not occur naturally (e. g. Cogo and Dewey 2006; Jenkins et al. 2011). Several researchers have explored ELF interaction by creating contexts for ELF users to communicate, whether in targeted discussions or casual conversations (see House 2003; Jenkins 2000; Lesznyák 2002). Although such communication elicited by researchers is not naturalistic, findings from naturalistic ELF interaction are not inevitably applicable to the majority of naturalistic ELF interactions either. As mentioned above, Jenks (2012) noted the importance of the communicative context and the accompanying norms, expectations and goals in shaping the appropriateness of interactive behaviour. Results from naturalistic ELF interaction in a particular communicative context may therefore be substantially different from findings from a different communicative context.

One of the benefits of researcher-elicited communication is that the design of the communication task can influence some of the above-mentioned expectations, goals or other cognitions from interlocutors; researchers can then analyse how ELF interaction and interlocutors’ expectations and goals for communication might affect each other. An additional benefit to researcher-elicited communication is that ELF interlocutors’ norms, expectations and goals for particular interactions can often be explored in a more direct manner than is typical in research on naturalistic interaction. This includes the use of stimulated recall. Researcher-elicited communication thus cannot substitute for naturalistic interaction, but can focus by design on particular aspects of interaction.

3.1 Participants

The four participants targeted here were from a group of 40 ELF users who were students at an English-medium university in Montreal, Quebec, Canada. The official language of Quebec is French, but at the university and in the multilingual and multicultural city of Montreal, students had opportunities to hear and use various languages, including their L1s. However, English was the language of instruction at the university. Participants were recruited through notices (in English) posted throughout the university campus inviting them to participate in speaking tasks, with compensation. All participants self-reported L1s other than English. Before, during and after the speaking tasks, participants were not directed to use English but used it as the predominant language of communication at the university and as their communicative medium of choice with unfamiliar interlocutors in the university setting. Each participant engaged in different researcher-designed interactive tasks with a previously unknown partner of a different L1. The particular task analysed in this study was a map task (described below). The participants in this study made up two pairs, selected for analysis because transcription of their map task interactions had shown extensive negotiation of meaning. One pair had achieved the shared goal of the task and the other pair had not fully achieved this goal, which allowed the chance to examine how interlocutors’ use of CSs and their perceptions of interaction were related to goal achievement.

One pair consisted of two male doctoral students in engineering: Jun Yi and Hamal (all names are pseudonyms). The second pair was made up of Jing, a female MA student in Business Administration, and Ajit, a male MA student in electrical engineering. Participants had resided in Montreal for a mean of 1.6 years (0.5–3). Their academic programs had different admission requirements for English proficiency scores. However, all four had received at minimum an overall score of 85 for TOEFL iBT or 6.4 for the IELTS proficiency tests, which was deemed sufficient to pursue graduate academic degrees. Prior to completing the task, participants also self-rated their English ability in listening and speaking on a nine-point Likert scale (1 = low, 9 = high). To summarize, in each pair the interlocutors were at the same level of graduate studies and had had similar proficiency scores and self-ratings for listening and speaking, although Jing’s IELTS score for listening was slightly higher than her partner’s. Table 1 summarizes participant information.

3.2 Procedure

The participants read and completed the consent form and a questionnaire about their biographical information and language learning history and abilities. In the first phase of data collection, they were paired and completed four interactive speaking tasks, each preceded by written and spoken instructions. The tasks were carried out in a quiet room and were recorded onto a laptop computer using a digital video camera and two lapel-mounted wireless microphones. The first task was an interactive warm-up task to allow participants to become more comfortable and familiar with each other. Participants then engaged in three speaking tasks. The map task (the second of the three tasks) is the focus of this analysis.

The map task was an information-exchange task, where both participants were required to exchange information that was unknown to their partner in order to achieve a specific, shared goal (Ellis 2003). Each participant had one of two versions of a map. One version showed all ten of the landmarks on the map, but did not have a route. The other version showed a route and six of the ten landmarks. The participants faced each other across a table, with a barrier blocking a view of their partner’s map. They were told orally and in writing that they each had an incomplete map: one participants’ map had the complete route, while the other’s map had all the landmarks. They were instructed to describe their maps to each other in order to accurately add the missing information and finish with two similar, complete maps (the goal). They had seven minutes to do the task.

The next phase of data collection began within 15 minutes of completing the map task. This phase elicited participants’ recall and perceptions of the interaction during the four tasks. Participants went to separate rooms to do a stimulated recall task, where each participant worked individually with a research assistant (RA) who had been present during the interactive tasks but had not previewed the videos before the stimulated recall. Before replaying the task videos on a computer, the RA explained the main focus of the stimulated recall, which was the participant’s memories of instances when at least one of the partners was faced with potential or actual difficulties in understanding or in being understood. The participant was encouraged to use the computer mouse to stop the recording at any point to tell the RA what the participant was thinking. The participant was also encouraged to stop the recording at any time when, during the task, he or she had noted (a) speech that was difficult to understand or (b) an attempt to prevent or repair difficulties in understanding experienced either by the participant or by his or her partner. The participant was encouraged to describe his or her thoughts at that moment in the original task. In order to become familiar with the demands of the stimulated recall task, the participant first viewed the warm-up task. The stimulated recall task then continued, with the video of the map task being shown to the participant. The entire stimulated recall session was audio-recorded. The RAs also took notes about participants’ comments and their timing in relation to the original task.

4.1 Types of strategies used

The first research question asked what types of CSs were used to address potential or actual difficulties in understanding in interactions between ELF users. Table 3 shows the frequency of strategy use by each interlocutor. Overall, both pairs used similar strategy types. However, their distribution varied greatly within and between pairs. For example, in Hamal and Jun Yi’s pair, Hamal would repeat what Jun Yi had just said, and quite frequently asked Jun Yi to clarify meaning or to confirm Hamal’s understanding. Jun Yi used these strategies much less frequently, but he would check to make sure Hamal had understood his meaning and would rephrase his utterances.

In Ajit and Jing’s pair, Ajit asked Jing to clarify meaning or to confirm Ajit’s understanding. However, Jing tended not to ask for clarification but only for confirmation that she had understood correctly. Ajit tended to repeat Jing’s utterances more than the converse. Jing was the only one of the pair to elaborate on her utterances, and she also tended to give summaries of her understanding to that point. The one strategy type which was not shared between pairs was the use of gesturing or visual aids. Both pairs had been instructed by the RA not to show their maps or to point. However, Jun Yi and to a lesser extent Hamal frequently made gestures above the barrier in the middle of the table while completing the task. Thus, both pairs used similar types of strategies when addressing potential or actual difficulties in understanding, but differed from each other in the frequency with which they used a given strategy during the task.

4.2 Strategy use when shared goal is not achieved (Jun Yi and Hamal)

The second research question asked how the ELF users who achieved or did not achieve a shared goal used these strategies and how they interpreted this use. The first pair of participants (Jun Yi–Hamal) did not achieve the shared goal of producing two similar, complete maps. The nature of their CS use is demonstrated in the way that Jun Yi began the map task (Excerpt [1]). He had a version of the map with a complete route, but with missing landmarks. He started by using spatial references which required a shared unit of measurement and orientation to an x and y axis, including a route which he described in terms of degrees from the x axis (see the appendix for transcription notations used in the excerpts of interaction).

From the beginning, Jun Yi took charge of how the task would be conducted. He started to use gestures almost from his first utterance, and introduced and implemented his approach towards spatial references, using them to describe the route to Hamal. When Jun Yi and Hamal separately viewed this part of the interaction in the stimulated recall, they both commented on the use of this system for spatial reference.

The pair continued to use this spatial reference system for another minute and a half, with Jun Yi providing information about the route and Hamal frequently asking for clarification or confirming his understanding. However, Hamal then attempted to provide Jun Yi with some information about a landmark.

Three points stand out in this excerpt. First, Jun Yi was determined to complete the task using a particular process, even ignoring Hamal’s contributions or demanding that Hamal wait until Jun Yi had finished describing his own map. Second, both interlocutors, especially Jun Yi, used gestures quite frequently. Finally, although Hamal tried twice to describe part of his map, when he was ignored or told to wait, he did not show any negative reaction. He continued, asking for clarification or confirmation in order to accurately reproduce the route Jun Yi was describing. Although both interlocutors viewed this part of the task in the stimulated recall, only Hamal made any comments about it.

At this point in the stimulated recall, Hamal attributed the lack of task completion to Jun Yi’s insistence that the missing route be completed before describing the missing landmarks. However, Hamal did not seem to be upset by this insistence. After this episode in the map task, there was further negotiation of meaning about the missing route. In the stimulated recall, Hamal commented on two periods, 40 seconds and 60 seconds long, respectively, where much negotiation was occurring about a precise direction.

Jun Yi eventually stated that he was finished describing the route and requested that Hamal describe the missing landmarks, shown in Excerpt (7). At this point, six and a half minutes had passed since the task had started.

Jun Yi begins the description of landmarks by starting to give Hamal instructions. However, Hamal interrupts Jun Yi and starts describing a landmark (three trees) in relation to the points on the route that Jun Yi had described. Jun Yi requests clarification and confirmation to ensure that he has correctly understood what the landmark is. His confusion may be due to Hamal’s consistent pronunciation of three trees with a flapped r/θɾi tɾiz/. Jun Yi asks Hamal twice for information and clarification about how the trees are spaced in relation to each other, with Hamal making many gestures while he describes the trees’ location. In the stimulated recall, only Jun Yi made any comments about this period.

In fact, the last utterance by Hamal in Excerpt (7) was the final utterance in the task, since the seven-minute time limit had expired. Jun Yi made a final comment in the stimulated recall about the task overall.

4.3 Strategy use when shared goal is achieved (Jing and Ajit)

The other pair of speakers (Jing–Ajit) completed the map task within the allotted time. They used similar strategies as the first pair, but their strategy use showed a different pattern of interaction. Like the first pair, Jing and Ajit engaged in much negotiation of meaning. The use of strategies to address difficulties in understanding can be clearly seen in Excerpt (10), when Jing was trying to describe a landmark and its location to Ajit, who had difficulty understanding the nature and position of the landmark. Jing had a version of the map with a missing route, but with a complete set of landmarks.

Jing initially used an unusual choice of words (“stop mark”) to describe the landmark. Ajit requested clarification of stop mark, and Jing rephrased to use the more standard term stop sign. Ajit also had difficulty understanding the phrase of location “right above”, which he initially seemed to believe meant “on the right and above”. With Ajit requesting confirmation and clarification of the location, Jing repeated and rephrased her utterances and gave additional information about the landmark’s position until both speakers were satisfied that the position was understood. Both speakers commented on this period in the stimulated recall.

The stimulated recalls show, first, that Jing believed that she had clearly communicated the location of the stop sign. However, she realized that, at least initially, Ajit was not sure where the stop sign was located. In her second utterance describing this period in the stimulated recall, Jing stated that Ajit had difficulty figuring out whether the stop sign was on the left or right. In Ajit’s stimulated recall, he repeated verbatim the phrase “right above”, but stated that Jing had initially given incorrect information about whether the stop sign was on the right or the left.

Due to space considerations, additional excerpts from Ajit and Jing’s map task cannot be shown. However, during the task, Jing used the phrase “right above” or “right below” three more times to refer to landmarks. Each time, Ajit signalled difficulty in understanding and each time, the pair negotiated meaning until they both believed the location of the particular landmark was understood. It is clear from both stimulated recalls that neither speaker realized the source of difficulty in understanding. Jing did not realize that Ajit understood “right above” as “on the right and above,” while Ajit did not realize that “right above” meant “immediately above,” not “on the right and above.” Despite this difficulty, Jing and Ajit completed the task, with Jing accurately recreating the missing route and Ajit accurately placing all but one of the landmarks, the stop sign described in the excerpt, which he placed above and to the right of the house. It is notable, though, that the other three landmarks which Jing described as “right above” or “right below” another landmark, were accurately placed by Ajit.

5.1 Strategy use and interaction dynamics

The CSs used by both pairs were generally similar in type to those observed in previous research on ELF users, such as repetition or clarification requests to signal difficulties in understanding, and elaboration and rephrasing to address the difficulties (e. g. Mauranen 2006; Matsumoto 2011). None of the ELF users in the current study drew on the “let it pass” strategy (Firth 1996). This is likely because to achieve the shared goal of two complete, similar maps, the interlocutors needed to accurately exchange information, so difficulties in understanding had to be signalled and addressed. In other words, the design of the communication task encouraged listeners to engage in close listening and accommodation to their interlocutor, by repeating the interlocutors’ words or phrases (Cogo and Dewey 2006). Similar to House (2003), the interaction examined in this study was not naturalistic; it was elicited using a researcher-provided communication task with interlocutors unfamiliar with each other. The communication task probably elicited greater use of strategies to address difficulties in understanding than some naturalistic interactions. However, I would argue that in working towards the shared goal, the ELF users in the current study showed an unmistakable focus on meaning, not language, and were using ELF in authentic communication.

In terms of how CSs were used and interpreted by these ELF users, the excerpts and the stimulated recalls show clear differences both within and between the pairs. These differences were likely related to what has been referred to as participant orientation or patterns of pair or dyadic interaction (Kim and McDonough 2008; Storch 2002). Throughout Jun Yi and Hamal’s interaction, Jun Yi consistently strove to manage the interaction. For example, when discussing the route, Jun Yi first set a common reference for measurement and spatial orientation, then used that reference to describe the route on the map, sometimes checking Hamal’s comprehension. When Hamal did attempt to tell Jun Yi about the location of a landmark, Jun Yi first ignored him, then told him to complete the route first. When the landmarks were being discussed, Jun Yi initially attempted to tell Hamal how to complete that part of the task. In other words, whether Jun Yi possessed or whether he required missing information, he consistently tried to manage the interaction and to shape Hamal’s contributions to the interaction. For the most part, Hamal did not challenge this pattern of interaction and usually ceded control of the interaction to his partner. Researchers have termed this pattern of interaction dominant/passive (Storch 2002) or asymmetric (Galaczi 2008).

The distribution of CSs in Jing and Ajit’s interaction, shown in Table 3, reflected another pattern. Generally, Jing assumed responsibility for the rhythm of the interaction. She used various CSs to ensure that she and Ajit understood each other, by asking for confirmation of her understanding of an utterance, elaborating on previous information when Ajit signaled difficulty understanding, and summarizing what had been said or what she understood to that point. However, Excerpt (10) shows that even when she was prepared to go to the next item (so the road now), she was attentive to Ajit’s move to continue discussing the previous item (On the right OK). Jing’s openness to Ajit’s contributions was apparent throughout the transcript. Though her speech often overlapped with that of Ajit, Jing rarely claimed a turn before Ajit had clearly finished or relinquished his turn. Researchers have called this pattern of interaction collaborative (Galaczi 2008) or expert/novice (Storch 2002), where one interlocutor (expert) has more control over the interaction but both interlocutors show high participation and engagement with each other’s contributions.

5.2 Strategy use as a shared responsibility

In research on L2 acquisition, studies on adult learners’ patterns of interaction have often explored how paired interaction might be shaped by differences in L2 proficiency and how different interaction patterns might influence potential L2 development. Findings have usually shown that, regardless of proficiency, learners who interact collaboratively show more transfer of knowledge than learners whose interaction is not collaborative (e. g. Storch 2002). However, Kim and McDonough (2008) found that some L2 learners who had been dominant or collaborative when interacting with an intermediate interlocutor became collaborative or passive, respectively, with an advanced interlocutor. This difference may have been due to learners’ perception that their interlocutors were of higher language proficiency.

These descriptions of L2 learner interaction patterns are relevant for ELF research because, as Jenks (2012: 403) notes, “ELF research must expand its empirical database in order to understand the varied ways in which ELF interactions are managed.” The communication in the current study was not naturalistic, but was authentic in that interlocutors focused on accomplishing the task goal and almost never focused on language form in itself. That is, language proficiency differences and language learning were not made relevant. ^[3] What was relevant, however, concerned patterns of interaction. The ELF users in this study had to communicate about direction and distance accurately and efficiently to successfully achieve the shared goal of two complete, similar maps. In one pair, one interlocutor’s use of CSs seems to have been affected by his perception of his proficiency in the task; this perception also influenced the pattern of interaction and the explanations made about task success.

In the video recording, an RA gave instructions just before Jun Yi and Hamal’s map task. Jun Yi told the RA with a smile, “I think this task is too easy for me. My research is about computer graphics,” then addressing Hamal, “Leave this one to me.” Given these statements, Jun Yi’s subsequent behaviour in managing and generally dominating the interaction during the task is not surprising. Although Jun Yi seemed to believe that he was especially suited to do this task and did communicate his information with a high level of precision, the joint goal was not achieved. Both Jun Yi and Hamal believed that the other’s behaviour was the reason for not achieving the goal, and in Jun Yi’s case, his behaviour may have been due to his self-perceived high proficiency in doing tasks of this kind.

The second pair, Jing and Ajit, did not make any explicit references to their proficiency in doing the communication task. However, they achieved the shared goal. This was despite the fact that neither was aware that they had different understandings of the phrases right above and right below. After using “right above” or “right below” with three other landmarks, Jing and Ajit spent a total of 16 turns discussing their locations, and never explicitly showed any frustration either verbally or visually, although in the stimulated recall both stated that for these items, their interlocutor had been either unclear or strangely uncomprehending. Neither explained why they had appeared so composed during these episodes, but Jing’s comment from an earlier stimulated recall may be significant.

I think that’s the two reasons I don’t understand people, like most of the people from different places. If they speak slower … if their accent is okay – you may have a British accent. You may have a Bangladesh accent, but … as long as you speak slowly, clearly, I can understand. That’s one thing that I find out through the years that I have been studying English. So that’s why I’m trying my best to speak with people like slowly and clearly too [italics added].

Jing demonstrates her view of communication as a shared responsibility, where all involved interlocutors should feel an obligation to make communication successful (see also Firth’s 2009 study on Business ELF users). This view is consistent with a collaborative pattern of interaction where, as Storch (2002: 128) notes, “a pair work[s] together on all parts of the task … [and] alternative views are offered and discussed, leading to resolutions that seem acceptable to all participants.” ^[4] It seems that the effect of sharing linguistic or non-linguistic resources is enhanced when all interlocutors’ contributions are considered potentially helpful.