Open Research Online Designing and Supporting Virtual Exchange: The Case of Chinese–English e-Tandem

this article examines how adult distance learners with limited levels of second language (L2) proficiency can engage in and benefit from e-tandem learning and to what extent intercultural learning takes place in the context of a six-week guided e-tandem project. While uK participants have completed a one-year beginners’ Chinese course at the open university, their proficiency level in Chinese remains modest (between A1 and A2 on the Common european Framework of reference scale). their Chinese counterparts are undergraduates at Beijing Foreign studies university with eight years of studying english behind them and majoring in english. this asymmetry presents particular challenges. through analysis of email exchanges, transcripts of skype conversations and interviews, surveys, and learning diaries, we research learner engagement and development as well as identifying areas where the design of the exchange might be improved and support for less advanced learners strengthened. our findings suggest that while there is significant evidence of L2 development and intercultural learning, this can come at a high price for learners. our conclusion, based on participant feedback, outlines modifications to the standard e-tandem model which might provide more effective support for those who are struggling and make for a less challenging but equally enriching experience. these findings have pedagogical implications for the future design of e-tandem projects, in particular how to best support e-tandem exchanges involving asymmetry, or where participants have limited L2 proficiency.


introDuction
english-speaking learners of Chinese face significant difficulties in memorizing large amounts of characters containing lexical information, in dealing with unfamiliar syntax, in learning romanized pinyin form and character script, and above all in mastering tone-based pronunciation (Kan,owen,. distance learners at the open university (ou), many of whom live outside large conurbations, will have little opportunity for authentic interaction with competent speakers of Chinese and relatively few opportunities for speaking practice of any kind. using virtual exchange is one possible means of providing these.
until recently most virtual exchanges took place either within europe, or between europe and north America, and were devoted to communication in the romance or germanic languages (including english) (Lewis and o'dowd 25). the past decade, however, has seen a significant growth in the number of virtual exchanges between europe and China, as reported in a range of publications (e.g. Kan,stickler,. of these exchanges, three employed an e-tandem or teletandem model (see below). Kan, stickler, and Zhu concluded that the perceived benefits of e-tandem learning include: practising colloquial expressions; learning about aspects of culture; increased personal confidence; the satisfaction of helping partners; and taking part in authentic communication (135). Jin, for her part, found that while the Mandarin Chinese produced by American learners increased in quantity in the course of a ten-week exchange, gains in quality were less evident. she attributed this to a reluctance by Chinese partners to provide scaffolding in the form of error correction, as to them this would have seemed impolite (208-9). Cappellini studied the development of L2 interactional competence by analysing conversational side-sequences to chart the dynamic relationship between expertise roles and the provision of linguistic scaffolding. in doing so, he demonstrated that in e-tandem learning primacy is given to topic explanation and to lexical acquisition sequences (10)(11)16). ruan and Medwell focused on the use of social networking tools installed on mobile devices to develop L2 proficiency and intercultural communicative competence in a group (n=9) of secondary school gCse-level learners. their students reported that using sns technology, which promoted social presence, they developed writing and reading skills more than speaking and listening skills (374).
As well as offering opportunities for interaction, virtual exchanges can also present significant challenges. in the context of our study these include contrasting expectations, limited levels of L2 proficiency, a large time difference, disparate technological and educational cultures, to name only the most obvious. this study explores the challenges encountered in undertaking an inter-university e-tandem exchange between learners of Chinese and english, examines how they can be surmounted, and charts the extent to which participants nonetheless benefit from such an exchange. it draws on a range of evidence, including programme documents, participant testimony, and recorded interactions.

literAture review 2.1 tAnDeM leArning
tandem learning is a versatile and well-established method of language learning, based on mutual assistance by peers, provided within the framework of a cross-cultural partnership. While its guiding principles are autonomy and reciprocity, other key features of tandem learning include 50/50 dual-language use and error correction by competent L2 speaker partners. partners are tasked to act as both linguistic and cultural informants for each other (though not as actual teachers). in other words, a tandem exchange habitually combines intercultural dialogue with an equally dialogic approach to language learning. e-tandem learning normally combines the use of multiple (i.e. textual, aural, and visual) communicative modes and frequently of more than one medium (Lewis and peters 11).

AutonoMy
As david Little points out, "there are two quite distinct interpretations of what it is to be an autonomous learner. While the first of these emphasizes the 'individual-cognitive-organizational aspect of learning', another tradition sees autonomous learners as 'working together … taking 3 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 on quite a complex interactive life as learners'" (Little and thorne 17). tandem learning combines both aspects of autonomy, operating in the intersubjective cognitive space identified by Vygotsky as the Zone of proximal development (Vygotsky,Mind in Society 86). here learning in partnership with a more knowledgeable other leads ultimately to individual growth: "the goal of development is behavioural autonomy, the ability to do things for oneself; and progress through each successive developmental zone is a matter of using the support of others to build on the autonomy already achieved" (Little,dam,and Legenhausen 48). given that tandem learning is based on partnership, autonomy in the tandem context also involves respecting and contributing to the autonomy of one's partner (Lewis,"theory of sociality" 56) since it is an established principle of tandem learning, the present project depended on participants being able to display a degree of learner autonomy. And since participants were mature and motivated, it might have been assumed that they were perfectly able to manage their own learning. Yet recent years have seen considerable discussion of the extent to which learners with lower levels of L2 proficiency are capable of exercising autonomy. thanasoulas, Zhang and Li, deng, sakai and takagi, and Valadi and rashadi all found a significant correlation between L2 proficiency levels and learner autonomy. More recently, however, a quantitative study, by Zarei and Zarei found that "there were no significant differences in the autonomy of eFL learners at different proficiency levels" (268), while Ünal, Çeliköz, and sari confirmed that "language proficiency was not an influential factor in students' motivation and autonomy" (121). in the face of contradictory evidence, and keenly aware of the limited level of L2 proficiency of ou learners, the project organizers decided to provide higher levels of scaffolding (and less scope for the exercise of autonomy) than might be expected in a tandem exchange. in particular, rather than having absolute freedom of discussion, learners were assigned six weekly topics. For each of these a worksheet was devised, providing L2 language support.

reciprocity
Brammerts defines reciprocity in this context as a case where "both partners should contribute equally to their work together and benefit to the same extent" and that "learners should be prepared and able to do as much for their partner as they themselves expect from their partner" ("tandem language learning" 11). the motivational power of reciprocity is attested by the fact that it is one of the key features of human sociality and a necessary condition for human cooperation (Camerer and Fehr 56;Fehr and Fischbacher 785). in relation to tandem learning, it has been asserted that "a tandem partnership … will only last if both partners benefit from it (preferably both to the same extent)" (Brammerts,"Autonomous Language Learning" 31). in other words, "the provision of support for the partner is a prerequisite for being able to expect support from them" (31).
in tandem learning, though, reciprocity is not simply a matter of gaining equal benefit from an exchange; it is also about providing mutual support. Although tandem partners are not teachers, their role is to help each other learn. "Within a tandem partnership, partners support each other in their learning, i.e. they will offer the assistance they have been asked for to the best of their ability. Both correct each other, suggest alternative formulations, help with the understanding of texts, translate, explain meanings, etc." (Brammerts,"Autonomous Language Learning" 32).
scholars such as Koch have interrogated the principle of reciprocity, observing that "social reciprocal behaviour [is] a distinctive or common feature of human language and culture in general" (Koch 129). But as Brammerts points out, the mutual benefit implied by reciprocity does not equate to identical benefit: "tandem partners may -and often do -have different goals. … tandem work is very suited to the collaboration of partners with different learning habits and needs, and even different linguistic levels" ("Autonomous Language Learning" 34). While it is helpful for e-tandem partners to have much in common, including equivalent levels of L2 proficiency, this can rarely be guaranteed. nor is it entirely necessary. "Learners with very little knowledge of the foreign language may very well make suitable partners for advanced learners, since they will have a better command of their mother tongue than their learning partner" (34). in reality there is likely to be a degree of asymmetry in any tandem learning partnership, whether in terms of objectives or of learning habits. 4 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 in this particular project, participants were demographically well matched. Both groups consisted of university-based mature adult distance learners, with a similar minimum age (26 years for ou participants; 27 years for those from Beijing Foreign studies university (BFsu)). Learners in both groups combined the pursuit of a professional career with continuing academic study, which implies that they were self-disciplined and strongly motivated. But different patterns of study meant that their experience of second language learning and their respective levels of L2 proficiency were very different. the ou students had begun learning Chinese on entry to the university and had benefited from just one year of study, while their BFsu counterparts had been studying english (initially at secondary school) for up to eight years. this meant that while the L2 proficiency level of the ou group was somewhere between A1 and A2, that of their BFsu counterparts was at least B1. this asymmetry was involuntary but unavoidable, since no other groups were available.

DuAl lAnguAge use
one of the conventions of tandem learning is that equal emphasis is placed on the use by each partner of the two languages featured in an exchange. in spoken tandem learning, this means that each partner owns half of the meeting, during which time their needs are addressed mainly in their target language. hence, regarding text-based tandem, Brammerts states: "When writing to one's partner … one writes half of the time in the mother tongue, producing the necessary linguistic input for one's partner. the other half of the time is spent … writing in the foreign language" ("Autonomous Language Learning" 32).

tAnDeM leArning As interculturAl DiAlogue
tandem learning has long claimed to be a form of intercultural learning. that claim rests essentially on its dialogic nature. dialogic interaction as an approach to learning has an illustrious pedigree: its origins reach as far back as socrates, while more recent proponents include Bakhtin, Freire, habermas, and sen. intercultural dialogue has been adopted as policy both by the Council of europe and by unesCo, which created an e-platform to promote it in 2018. it has been defined as follows: intercultural dialogue is a process that comprises an open and respectful exchange or interaction between individuals, groups and organisations with different cultural backgrounds or world views. Among its aims are: to develop a deeper understanding of diverse perspectives and practices; to increase participation and the freedom and ability to make choices; to foster equality; and to enhance creative processes. (Council of europe Living together 10) Aviva doron, a unesCo Chair in intercultural and interreligious dialogue, stresses that dialogue entails listening, eschewing positionality, and allowing oneself to be stretched: dialogue distinguishes itself from debate because it involves a form of listening that is beyond position or profession. dialogue is described as a process of exchanging information where participants leave the dialogue with a deeper knowledge level and wider frame of reference than when they approached it. it involves the creation of an expandable context. dialogue is successful to the extent that all parties to it are permanently stretched beyond their opening views. (doron) doron also stresses the interpersonal nature of dialogue: "a dialogue may request from its participants to see each other not as an abstract being, but as a particular individual and the process as one of accepting the other". tandem learning is based on the conviction that interpersonal dialogue is a singularly effective approach to intercultural learning. the aim is to avoid some of the more obvious kinds of 'groupthink' that occasionally afflict virtual exchanges based on collective, class-based interactions.
the process of dialogue is not always comfortable and offers no guaranteed outcomes. robert heath and his coauthors observe that "dialogue is problematic because truly hearing the voice of the other involves a struggle against many obstacles" (heath et al. 361). they also affirm that dialogue "produces unintended and unpredictable outcomes" (366). Among the outcomes that can be hoped for are "a relational resolution that develops from understanding each other's emotions, values, interests, and positions" (367-8), but it is equally possible that there will be no actual resolution of differences or problems. 5 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 What participants in tandem learning can aspire to is "understanding … or personal transformation" (heath et al. 389). Conceiving dialogue as a skill-based process, the hoped-for personal outcome is what Michael Byram conceives of as "the intercultural speaker; someone who draws expertly from skill repertoires in mediation and conflict resolution, is curious and open towards others, and is critically cognizant of and explicit about his/her own values as well as those of the other" (qtd. in ganesh and holmes 83). Viewing the process in terms of psychology rather than skills, doron offers instead the concept of "the intercultural person … whose cognitive, affective and behavioural characteristics are not limited, but are open to growth beyond the psychological parameters of any one culture" (doron). this, she claims, is more inclusive than other possible designations, in that it denotes "the type of person whose psychological makeup transcends any type of group identity including national, racial, ethnic, gender, professional, or other sociological typifications".
Finally, the role of dialogue in learning finds theoretical support in both the linguistic philosophy of Mikhail Bakhtin and the developmental psychology of Lev Vygotsky. in the words of the Canadian educationalist gordon Wells (a disciple of both), it has led to "the dialogic turn" (Wells 268) in late twentieth-and twenty-first-century education. their "insights, taken together, make a compelling case for reconceptualizing learning and teaching in terms of a dialogue of collaborative knowledge-building" (269). Moreover, as Wells explains, "the dialogue i have with others can be internalized and the search for understanding continues in the dialogue of what Vygotsky called 'inner speech'" (269).
this offers a clear justification for adopting an approach to learning based on a series of dialogic interactions between individuals, as embodied by e-tandem. though there is no certainty about the level of intercultural competence each partner will attain, especially given the relatively short timeframe of an exchange, e-tandem at least offers them an opportunity to set foot on the pathway to intercultural personhood.
taking part in such an exchange is not, however, risk free. For ou learners, who study predominantly at home and alone, building a successful partnership with a distant member of another culture is not automatic. More importantly, Brammerts points out that "tandem language learning requires a certain minimum proficiency in the target language" ("tandem language learning" 30). A worst-case scenario includes the possibility that those with lower levels of L2 proficiency might become overstretched and demotivated.

reseArch Questions
With these considerations in mind, the present study seeks to address the following two research questions: • to what extent can learners of limited and unequal levels of L2 proficiency engage in and benefit from e-tandem learning? • to what extent do e-tandem exchanges promote online intercultural learning? By addressing the above questions, the study aims to establish some key design principles to best support e-tandem exchange between partners at limited and unequal levels of L2 proficiency.

reseArch context
the beginners' Chinese course studied by ou participants is a thirty-credit module. students can study it at level 1 as one of the options in a chosen degree pathway, or as a standalone course. While most students are British or eu nationals based in the uK, a small proportion is of different nationalities based in other countries. the course, designed for part-time study across thirty-seven study-weeks, teaches Mandarin Chinese and simplified characters. digital skills such as how to word process characters are taught at the start of the course. By the end, students are expected to recognize and type about 500 characters and achieve A2 level as defined by the Common european Framework of reference for Languages (CeFr) (Council of europe Common European Framework). however, due to the unique challenges of learning Chinese (see introduction), the desired outcome at the end of the thirty-seven study-weeks is between CeFr A1+ and A2.
BFsu participants are drawn from the post-diploma BA english programme at its institute of online education. the programme is designed for those who already hold a diploma in english and the entry requirement for english is CeFr B1+, which means that they have studied english for at least eight years (six years in secondary school plus two years' diploma). it is a three-year programme with two semesters per academic year. By the end of the programme, students are expected to achieve CeFr C1. students who enrol are all Chinese nationals based in China.

the e-tAnDeM project
students from both universities were recruited on a voluntary basis to start this six-week Chinese-english e-tandem project at the end of october 2018. in each of the six weeks, participants were encouraged to i) write a short piece on a recommended topic using both Chinese and english (as per tandem conventions) and correct each other's written work (this was sent as email attachments); ii) have one skype conversation with each other for about thirty minutes, talking half of the time in Chinese and the other half in english; and iii) reflect in a weekly learning diary on the challenges and benefits.
prior to the start of the project, all participants received via email the project description explaining the two key principles of reciprocity and autonomy in tandem learning, and what participants were expected to do. they were also provided with the following documents: an informed consent form; a timetable; the pairing information; instructions on installing skype and the audio recording device 'Callnote'; a learning diary template; and most importantly, the weekly worksheets with the suggested topics and useful expressions (see Appendix A for a sample). the six topics are (in weekly order): getting to know each other, Family and friends, Food, Weekend, Cultural comparison, travel.

pArticipAnts
participants from the ou were recruited after they had completed a beginners' Chinese module. When BFsu students joined this project, they were in the second semester of their first year of study. the project began with twelve pairs but two pairs (pairs 1 and 9) dropped out in the first week, and pair 2 did not engage with a single task so they are excluded from the data. hence, the findings reported in this paper are based on nine pairs (eighteen participants). see table 1 for participants' profiles.
For anonymity, each participant is coded. ou stands for participants studying with ou, and BF for Chinese participants from BFsu. For example, the two participants in e-tandem pair 3 are coded ou3 for the ou participant and BF3 for the BFsu participant (see

DAtAsets AnD DAtA AnAlysis
Multiple datasets were compiled between october 2018 and February 2019 from the following sources.

Written work
A total of 157 files were collected. eighty-nine of these contained written messages on the six prescribed topics, of which thirty-eight were written by ou participants. A further sixtyeight were correction files (messages accompanied by amendments and annotations from L2 partners). the focus of this paper is not on feedback and correction, so they have been excluded from the current dataset. Messages written in Chinese have been translated by Author 2. translations follow the original in square brackets.

Skype conversations
twenty-two skype conversations were recorded using Callnote but only nineteen were transcribed because three of them were of poor quality and hence removed from the data. When Chinese was used, an english translation is provided in square brackets by Author 2.
learning diaries participants were encouraged to keep learning diaries as a means of recording their activities and reflecting on their experiences. Forty-six weekly learning diaries were collected, of which a dozen were by ou participants. Where the diary was in Chinese, an english translation by Author 2 follows in square brackets.

online survey and follow-up interviews
All the participants were invited, at the end of the project, to complete a follow-up online survey (see Appendix B) to evaluate and reflect on their learning. there were both multiple-choice and open-comment questions. For BFsu participants, all the questions were translated into Chinese for easy understanding and administered separately. Fourteen responses were returned, of which five were by ou participants.
Follow-up interviews were conducted with four ou students on skype to further explore how to best support learners with low language proficiency. All four interviews, each lasting about 30 minutes, were recorded and later transcribed.
Analysis our data was predominantly textual, which dictated a qualitative analytical approach. Written exchanges, recordings, interview transcripts, and survey open comments were compared and cross-referenced. our analysis focused on critical incidents, or "situations in which a misunderstanding, problem, or conflict arises as a result of the cultural differences of the interacting parties, or a problem of cross-cultural adaptation and communication" (Apedaile and schill 7). these are frequently seen as events that are out of the range of normal experience and therefore of heightened significance. in this context, they were simply salient points at which new information was being exchanged in such a way that learning of some kind appeared to be taking place.
in the next section, we present the main findings.

Challenges
there is little doubt that learners found the experience of e-tandem exchange challenging. As mentioned above, of the twelve pairs who began the exchange, two were unable to finish and one did not complete any tasks. of the two groups of learners, the ou participants clearly found the exchange more daunting, as the raw numerical data tends to confirm. of a total of eighty-8 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 nine emails sent by participants to their partners, thirty-eight were written by ou participants while fifty-one came from their BFsu counterparts. An unequivocal explanation of the disparity in participation rates is elusive, but it appears to have its source both in unequal L2 proficiency levels and a more directive approach to project management in Beijing. Columns 3 and 6 of table 2 reveal that while six of the BFsu participants completed all six weekly tasks, only three of the ou participants did so. no BFsu participant completed fewer than five written tasks, while three ou participants completed only two (see table 2 above).
Apart from the obvious difference in L2 proficiency levels, several factors were identified by ou learners as causing them difficulty. of the five ou participants who responded to a postproject online survey, all indicated that they found the spoken exchange in skype particularly problematical, owing to technical issues, which made the exchange inaudible. one ou participant reports, of the single attempt she made to use the medium: We tried to skype one time and the connection was so terrible that we could hardly hear anything each other was saying. it was quite -it was embarrassing, you know. [ou6, interview] As well as technical and audibility issues, participants also reported difficulties in fixing on a suitable time for a synchronous exchange, given the significant time difference (seven hours in summer, eight hours in winter) that exists between the uK and the prC. As one ou participant acknowledges, this probably disadvantages Chinese learners more, since they are likely to find themselves attempting an online exchange at the end of the working day, or when dealing with domestic responsibilities: usually, the time when our, sort of, timetables overlap tends to be in the morning, erm, which is not a problem for me, usually not a problem -But, for her, that time was just when she was putting the kids to bed and -it's, it was the busy part of the evening for her. so the scheduling was really difficult. [ou5, interview] ou participants reported other challenges associated with time management. some found themselves simply unable to budget for the amount of time required for a weekly exchange: i found that i didn't have time to give it, erm, the attention i thought it deserved. Because once a week was just too much and because i hadn't managed to do it once a week i got quite behind … so it got to week five and we hadn't even started …. er, she caught up by sending me everything at once and so then i felt overwhelmed. [ou6, interview] even those ou participants who persevered and completed the project remarked on the amount of time they had spent in preparing for their weekly exchange: For our 30-minute session, it was probably… Well, because i'm not working under pressure, i tend to take a long time to do everything nowadays. erm, i'd probably spend at least a couple of hours preparing it. And that would be, sort of, looking up sort of useful sentences that i could use, given -given the topic. [ou5, interview]

Benefits
Although taking part in e-tandem clearly stretched some ou participants to the limit, respondents to the online post-project survey remained clear about what they saw as its potential benefits: • interaction with a native speaker to enable learners to cope with real language; • practice and learning about other cultures; • the chance to improve one's speaking abilities; to improve language being studied; above all learning about each other's culture; • any exposure to a native speaker will be beneficial; • authentic interaction with real native language speakers.
Analysis of recorded interaction data and the testimony of participants bears this out and makes clear what were the tangible benefits of authentic interaction with native speakers. Writing practice and vocabulary expansion the first benefit was that participants gained practice in writing and learnt to use a significant number of new characters. this is illustrated by comparing the introductory message composed in Week 1 by ou7 with that sent by the same participant in Week 6 of the exchange. ou7's Week 1 message was very short; it contained ninety-four characters, of which many were repeated (e.g. 您, 我, 的). Most sentences contained grammatical (e.g. 您英文学有五年), lexical (e.g. 封信), and stylistic errors (e.g. 您). ou7 acknowledged that she had been obliged to use a dictionary to find every word (使 用字使 使 使 使 使 典 ou7's Week 6 email message (429 characters) is significantly longer than her first and contains fewer errors. A section of the first paragraph (123 characters) is reproduced below. ou7 does not know how to type one character so she uses pinyin (lu) instead of (旅). she uses a range of vocabulary (e.g. 旅行，游览， 去) and forms complex sentences using the correct linking words (e.g. 这个星期是我们最后的星期, 因此我们必须集中在话题, 但是钱是一问题使 使 使 使 使 使 ). she has also learnt how to type the Chinese full stop correctly, using a circle (which she had not used in Week 1). Although inaccuracies remain in roughly half of the sentences, the overall meaning can be viewed as well conveyed: [english translation: this is the last week of our e-tandem learning project. so we must focus on the topic of travel. i myself like travel but money and getting a visa can be a problem. i've been to the usA twice and Australia, Malaysia and trinidad once. i also toured some european countries but i have never been to China. i would like to go one day.] though lack of space prevents us from reproducing it extensively here, similar evidence is available from other participants who completed all six writing topics.

Multi-word and multi-character expressions
A second benefit, just as significant as the acquisition of new characters, is the extent to which learners from both institutions picked up the multi-character and multi-word expressions used by their partners. From the recorded interaction data, these included such homely phrases as 'sunday roast' and 'comfort food' (accompanied by an image of Yorkshire puddings), as well as 中药 [zhōng yào; traditional Chinese medicine] and 家庭团圆 [jiātíng tuányuán, family reunion].
Whereas in the conventional language classroom vocabulary tends to be learnt as individual items, a key feature of tandem learning interactions is that they are rich in the kind of common set phrases that are used in everyday conversation. the importance of this for second language acquisition cannot be overestimated. norbert schmitt reveals that, depending on the approach used, between 32% and 58% of both written and spoken discourse in english has been calculated to take the form of such conventionalized formulaic phrases, rather than individual words (schmitt 9-10). the proportion may be even higher in Chinese where more than 80% 10 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 of lexical items are made up of more than one character (duanmu). the prevalence and role of formulaic expressions in e-tandem learning has been highlighted elsewhere too  Learners testify that these formulaic expressions were not merely encountered and then forgotten, but internalized for recall and reuse: in her learning diary, ou6 complains about a poor skype connection and remarks: "i liked that i was able to teach her a translation for 断断续续 and it made her happy. otherwise, it's all her teaching me … We both learned a new phrase because of skype: 断断续续 Duànduànxùxù your voice comes and goes)." some six weeks later, when interviewed about this difficult experience, she recalled: We did manage to exchange -We both learned the word for "the sound comes and goes". so i said, "it's sporadic, it comes and goes", and [BF6] told me what that was in Chinese and, er, we laughed about that. that was the one thing we learned (laughs). [ou6, interview] Multi-word and multi-character expressions function on several levels. in terms of second language development, it has been convincingly argued that incorporating idioms and collocations into one's L2 repertoire helps learners develop both fluency and accuracy, because they constitute readily accessed "zones of safety", which learners can make rapid use of when producing the L2 in real-time conditions (Boers et al. 247).

Socialization and Pragmatic Competence
A further benefit of learning multi-word and multi-character expressions, such as those cited above, is that many such phrases, e.g. '中药' [traditional Chinese medicine], 家庭团圆 [family reunion], or 'sunday roast', are often culturally embedded-in which case "linguistic and cultural knowledge are constructed through each other" and "the learning of language, cultural meanings and social behaviour is experienced by the learner as a single continuous … process" (Watson-gegeo 339). in other words, the language in set phrases is so freighted with culture that learning one amounts to learning about the other.
Last but not least, it is contended that the "use of formulas is group-identifying" (Kecskes 71). thus, formulaic expressions "reflect the social behaviour of members of speech communities" and their use is part of "a community's shared language practices" (71). Consequently, gaining the ability to use them in an L2 context equates to developing pragmatic competence, which is both a linguistic and an intercultural skill. of course, neither the process of socialization into a second culture, nor the development of intercultural pragmatic competence, is completed within a six week period-that is not what is being claimed here. What can be seen, however, in the propensity for tandem learners to acquire new discourse sequences from their partners in the form of chunks, is the beginning of a process that has the potential to enable them to become interculturally competent L2 users, rather than merely L2 learners.
in answer to our research question one, therefore, it is clear that not all the learners who engaged in our e-tandem exchange became intercultural speakers, let alone intercultural persons. the ou participants in particular faced major challenges and some pairs were unable to complete the exchange. nonetheless, as seen above, those participants who did persevere, often in the face of technical, time-related, or linguistic difficulties, made demonstrable linguistic gains, 11 Lewis and Kan Modern Languages Open DOI: 10.3828/mlo.v0i0.372 which brought familiarity with cultural behaviours and enabled them to interact successfully with members of another culture.

rQ2 -to whAt extent Do e-tAnDeM exchAnges proMote online interculturAl leArning?
Any answer to this question must begin by making a distinction between cultural and intercultural learning. the separation between the two is not watertight. As the following discussion will show, acquiring the former can support the development of the latter. But still, these two kinds of learning are not identical; they are achieved in different ways and impact differently on the learner. While cultural knowledge is primarily cognitive and factual, intercultural competence involves what may seem a less tangible but more empathetic understanding of attitudes and behaviours on the part of the learner, which ultimately entails the appropriate modification of their own attitudes and behaviours.
in this project, one of the main instruments for encouraging participants to gain new cultural knowledge were the weekly worksheets. these were developed with a view to helping participants focus readily on a topic and initiate a conversation, using email and skype, as appropriate. it was of course hoped that this would lead to the reflective discussion of each other's cultures, but that is ultimately a process over which only the participants themselves have control. readers of this paper will be able to decide to what extent they feel this actually took place. data from skype recordings and interviews does suggest, however, that a considerable amount of intercultural learning took place-not necessarily in direct response to the prescribed topics, but in the kind of spontaneous interaction between partners that is part-and-parcel of authentic communication. this is evident in the variety of themes emerging from the data, some of which have little to do with the six weekly topics. Below, we first report the findings in relation to cultural learning, and then go on to address the extent to which intercultural learning may have taken place.

Cultural learning through discussion on food
Among the cultural learning reported by participants was knowledge about food, the family, the geography of China, its holidays and festivals, and Chinese traditions. All of these could simply be dismissed as the "four Fs" of culture: "food, fairs, folklore, and statistical facts" (qtd. in Zhu 21). But these students are learning part-time while, in many cases, holding down a job (witness ou7's earlier introduction of herself). Learning about cultures (and developing intercultural competence) has to begin somewhere and an everyday staple provides at least a starting point.
Food is indeed a staple of culture. As such, a worksheet was dedicated to it. in relation to food, ou5, for example, reported: For my partner, oats are considered to be a snack not breakfast. Chinese people usually have noodles and dumplings for breakfast. [ou5, Week 3 diary] ou5 may have been disconcerted by this news. he reciprocated by explaining the cultural significance of haggis, taking care to attach an image for his partner's inspection: haggis is a traditional food in scotland. it can be eaten at any time but there is one day each year when it is customary to eat it. this is the birthday of the famous scottish poet, robert Burns. on or around 25th of January people celebrate his birthday by eating haggis usually with turnips and potatoes for their evening meal.
[ou5, Week 3 written work] duly impressed, his Chinese partner (BF5) recorded in her diary for Week 3 that to "get to know a traditional scottish dish" was the aspect of that week's exchange she had enjoyed the most.

Geography
As a result of the following exchange, ou3 learnt about aspects of the geography of China as well as the correct names of two distinct but similar-sounding provinces. hebei, which means 'north of the (Yellow) river' and, farther west, hubei, which means 'north of (dongting) lake':  BF3：Beijing,shanghai,or shenzhen,um,you know,most, most boys will not leave home. they will live with their parents.
[pair 3, skype conversation -Week 2: Family and friends] For many Westerners, the best-known aspect of family life in China remains the 'one-child policy', adopted from 1979 to 2015. it is almost inevitable that this will form part of any discussion of the family. As can be seen in the following exchange, it was a topic for pair 3. the focus is on its impact on interpersonal relationships in the extended family. prompted by his interlocutor, the Chinese partner confirms that one impact of the policy is that cousins tend to be treated like siblings. in fact, the Chinese partner reacts by confirming this insight in a way that suggests he may be reflecting on it for the first time ('Yeah, yeah, that's true'), then offers linguistic confirmation of its accuracy ('Ah使 , cousin is like, it's called like, um, táng xiōng mèi'). [pair 3, skype conversation -Week 2: Family and friends] this interaction and the previous one demonstrate how linguistic and cultural knowledge are simultaneously negotiated in an e-tandem exchange. ou3 and BF3 acquaint each other with cultural phenomena (the 'empty nester', the 'stay-at-home son', and the 'cousin as sibling') which illustrate two rather different, but ultimately understandable, social dynamics. it is possible that sharing their very different understandings of family led to no further reflection on the topic by either learner. however, that seems unlikely for two reasons, which will be explored in the following section.

Discussion AnD conclusion
the ou participants in this study had limited levels of L2 proficiency, which were significantly lower than those of their Chinese e-tandem partners. taking part in an e-tandem exchange confronted them with sizeable challenges. despite this, the data presented above offer evidence of the acquisition of cultural knowledge and of a process of learning that focused not merely on facts, but also on behaviours, beliefs, values, and attitudes. it is not possible to demonstrate here that in six weeks, learners developed intercultural communicative competence. As Apedaile and schill point out, "developing [inter]cultural competence is a lifelong process" (44). nonetheless, participants' interactions offer clear evidence of an exchange of views on cultural phenomena such as attitudes to the dead or the concept of family that can hardly be dismissed as insignificant.
As Zhu hua reminds us, one longstanding and predominant metaphor for culture is that of the iceberg, in which values and beliefs frequently remain invisible to a casual observer, since what one actually sees is just the tip of the iceberg (188). introspective methods of data-gathering might have revealed more of what remained unspoken in our participants' conversations, but in the context of this study they were not available to us. this does not mean that our findings are negligible: to use a presumptuous analogy, astrophysicists regularly theorize about the nature of the universe, while conceding that a mere 5% of it is observable through human technologies.
A second reason for supposing that participants' spoken utterances represent only a part of their reflections can be found in Vygotsky's model of human development. For Vygotsky, ideas that are first shared in external, social speech are then pursued in the individual consciousness as 'inner speech', which he defines as thinking in pure meanings and characterizes as a dynamic, shifting, unstable thing, flickering between word and thought. in Vygotsky's own words, "while in external speech thought is embodied in words, in inner speech, words die as they bring forth thought" (Vygotsky,Thought and Language 249). in this sense, the dialogues studied in this article represent a kind of prelude to a more profound engagement with cultural meanings on the part of learners. inner speech is a widely acknowledged phenomenon, but also one about which many questions remain, especially in relation to L2 users (guerrero 2-3). Continuing our exploration into that realm was beyond the scope of our study and the resources at our disposal. nevertheless, our analysis demonstrates beyond reasonable doubt the potential of e-tandem learning as a means of launching the kind of dialogic thinking that leads to intercultural learning.

iMplicAtions
the purpose of this article is not simply to claim success, but also to consider how best to design and support e-tandem exchanges for learners of limited L2 proficiency. We have garnered many ideas-some in the shape of suggestions by our participants, while others have emerged from our own analysis. these are listed here for the benefit of readers whose prime interest lies in the pedagogical ramifications of our study. We shall then pass on to its implications for future research into e-tandem learning.

iMplicAtions For peDAgogy
Below are our main recommendations for the design of e-tandem projects for learners with limited and unequal levels of L2 proficiency: • Make the exchange a purely textual one, for an initial period, to allow partners to grow familiar with each other, before undertaking webconferencing sessions. • Make the programme less intensive by extending it over a longer period and alternating text-based exchanges (which are burdensome to prepare) with webconferences. • For less fluent learners, include simple practice activities, such as reading L2 texts aloud for partners to correct their pronunciation. • design worksheets to help participants find the common ground that benefits partnership (e.g. focusing on personal likes and preferences).
our own priority (based on scrutiny of the data) would be to find flexible, non-directive, ways of monitoring, supporting, and scaffolding participants in e-tandem exchanges. these could include: • guidance on how to provide error correction and specific, constructive feedback, without causing offence; • practising useful target language phrases, including gap-fillers, help requests, and polite ways of challenging partner's views; • managing expectations around spoken exchanges for those with limited L2 proficiency, and advising on ways of balancing the use of the two languages.
in relation to technology, skype has in recent years undergone a series of modifications which should have remedied some of problems reported by our learners. Fortunately, an increasing number of other webconferencing platforms are also now available. A possible alternative would be to use Ms teams or Zoom, both of which are accessible in China. it is vital that participants in virtual exchanges have multiple channels of communication-for instance, a text chat app, providing instant notification of messages, seems an increasingly useful tool.

iMplicAtions For reseArch
the present study relied heavily on the analysis of observable phenomena (written exchanges, speech recordings) to reach its conclusions. our initial focus was on the experience of e-tandem learning itself. in this regard, some of the core principles of tandem learning may be worth reexamining in the light of recent theoretical developments. there is scope for experimentation around modifications of these (e.g. the acceptance of translanguaging; new approaches to reciprocity for partners of unequal L2 proficiency).
our study applies the Vygotskyan model of cognitive development, and in analysing the data, we were persuaded of the explanatory potential of socio-constructivism. An attempt to probe this fully would have been conducted differently, had it been conceived in this way from the outset; it would certainly have involved the use of introspective methods to test the extent to which external dialogues engaged in with partners were then transformed into the dark matter of 'inner speech'.
one appropriate means of probing the contents of inner speech would be to employ stimulated recall interviews (sris), using as stimuli critical incidents occurring in the exchange. since sris need to follow closely on the incidents under investigation, such a study would require careful organization, including close coordination between both partners in the exchange. it is unlikely that such a project could be carried out (as the present study was) in the margins of one's teaching activity. A funded project, with research assistance and translation of data, is thus required. We are convinced that such a project would pay significant dividends.