Breaking the silence (again): on language learning and levels of fluency in ethnographic research

Ethnographic research is often multilingual, requiring the researcher to work in two or more different languages, if necessary with the assistance of an interpreter. Given this, surprisingly few ethnographers have attempted to discuss in detail how their own knowledge of different languages and their decisions to use interpreters and/or translators during fieldwork have affected the research they have conducted. Drawing on material from our own research, as well as from published accounts by other ethnographers, we aim in this article to dispel some of the ‘silence’ or ‘mystique’ surrounding such matters. More specifically, we argue for the importance of documenting and analysing not only the process of language learning in ethnographic research but also the ways in which levels of fluency in a second or additional language can affect the research process, including the writing of ethnographic fieldnotes and forms of self and other identification. We suggest that a heightened awareness of these issues can help researchers make more informed choices when carrying out and writing up ethnographic research using different languages.


Introduction
Ethnographic research is often multilingual, requiring the researcher to work in two or more different languages, if necessary with the assistance of an interpreter. Many researchers conduct fieldwork or interviews in one language (or more than one language) and then write up the results in a different one, drawing on academic research and other written sources published in these and possibly additional languages. For some, undertaking participant observation research involves another kind of ongoing work, namely learning a new language. As Karen O'Reilly has recently pointed out in her introduction to ethnographic research, this can be a complicated and time-consuming process: 'It may not simply be a matter of learning to communicate in another language, but of identifying subtle differences in dialect, understanding colloquialisms, acquiring slang terminology, and learning when and how to use a polite or a casual tone ' (2012: 95). In many cases, the possession of such skills by the researcher is necessary if they are to fully understand social interaction and the meanings attached to actions by participants in a particular setting.
Given the multilingual nature of much ethnographic work, researchers have written surprisingly little about the impact of language-related issues on the development of their own projects, from the initial research design stage right through to the dissemination of the results. More specifically, relatively few ethnographers have attempted to discuss in detail how their own knowledge (or lack of knowledge) of different languages and their decisions to use (or not to use) interpreters and/or translators during fieldwork have affected the research they have conducted. Almost twenty years ago, Bogusia Temple (1997: 607) argued that there was a 'remarkable silence' within sociology about matters relating to translation, interpretation and language skills in the research process. Not long afterwards, Axel Borchgrevink drew attention in a similar way to a 'silence' within social anthropology about issues of 'language competence' and the use of interpreters in fieldwork, a situation he related to the persistence of a kind of 'fieldwork mystique ' (2003: 95, 96). In the decade or so since then, a few ethnographers (notably, Tremlett, 2009;Rodgers, 2012) have provided detailed descriptions and analyses of their own language learning experiences before, during and after fieldwork, but such accounts still remain rare in the literature. 1 Against this background, our aim in the present article is to contribute to the development of a wider debate about key language-related issues in contemporary ethnographic research. More specifically, we are concerned to stimulate further discussion of some of the fundamental methodological and epistemological questions that arise when a researcher carries out fieldwork in a second or additional language. We begin by reviewing a series of attempts historically from within sociology and social anthropology 2 to break the silence on language learning and the use of interpreters in fieldwork, before briefly considering the extent to which professional codes of ethics, encyclopaedias and prominent textbooks on social research methods currently address these issues. Drawing on material from our own previous research, as well as from published accounts by other ethnographers, we then argue, more specifically, for the importance of documenting and analysing not only the process of language learning in ethnographic research but also the ways in which levels of fluency in a second or additional language can affect the research process, including the writing of ethnographic fieldnotes and forms of self and other identification. We suggest that a heightened awareness of these aspects of 'researching multilingually' (Holmes et al., 2013) can help researchers make more informed choices when carrying out and writing up ethnographic research using different languages.

Language learning, interpreters and the 'fieldwork mystique'
It is important to recognize at the outset that attempts have been made to stimulate discussion of language learning and the role of interpreters in ethnographic research ever since the emergence of sociology and social anthropology as academic disciplines. For example, the use of 'native languages' by field researchers was the subject of an interesting early exchange between Margaret Mead and Robert H. Lowie in the pages of the journal American Anthropologist (Mead, 1939;Lowie, 1940). 3 Some years later, Gerald Berreman (1972) published an account of how his research in an Indian village was affected by using two different interpreters, and this remains one of the most detailed discussions of the topic (although the author provides only the very briefest of comments on how he actually worked with the interpreters in question or attempted to learn the local language himself). 4 However, contributions of this sort have tended not to be followed by wider, more sustained consideration of the issues raised, but have instead usually remained relatively isolated interventions on the part of a few lone voices. Rather than the cumulative development of a body of disciplinary knowledge and reflection, the result has been a series of calls over the years by individual researchers, in much the same terms, for serious attention to be paid, finally, to such matters. Thus, it was still possible, in the early 1980s, for Ganath Obeyesekere to argue that 'the interpreter effect' was 'one of those problems we have swept under the carpet ' (1981: 11), and for Elizabeth Tonkin to note that 'anthropologists have often taken refuge in silence instead of thinking critically about how to improve language learning in the discipline ' (1984: 178). The failure of these important statements to provoke wider discussion of the issues among ethnographers in the years following their publication is indicated by the fact that almost two decades later other researchers felt able to advance strikingly similar claims about the absence of attention within sociology and social anthropology to language learning, the use of interpreters and translation. For example, Temple encouraged sociologists working with interpreters/translators to acknowledge the active role the latter play in social research and to debate conceptual issues with them, arguing that 'the figure of the interpreter/translator must come out from behind the shadows' (1997: 607, and see also Edwards, 1998). For his part, Borchgrevink (2003) criticized the lack of discussion within social anthropology of the use of interpreters in fieldwork and the noticeable failure to confront directly the question of the researcher's own proficiency in the language(s) spoken in the sites where they conduct their research.
How is the persistence of this general 'silence' about language learning and working with interpreters/translators in ethnographic research to be explained? As far as social anthropology is concerned, Borchgrevink suggests that among the factors contributing to such a situation are concerns about the researcher's authority (acknowledging language problems could potentially undermine their credibility) and 'romantic notions' or 'myths' about fieldwork (2003: 96, 115). On the second of these, he argues that: By making the practice of fieldwork the central and defining characteristic of the discipline, and indeed a precondition for the existence of anthropology, we place fieldwork outside the scope of serious critique. This creates the opportunity for mythologization, and results in what I have referred to as the 'fieldwork mystique'. (Borchgrevink, 2003: 114-115) According to Borchgrevink, the persistence of this view of fieldwork within the discipline results in key assumptions and practices escaping critical examination, notably those relating to language learning and working with interpreters. Remarking that many other aspects of fieldwork practice have now effectively been demystified, he concludes by emphasizing the need for 'an open discussion' of language skills, interpretation and translation in ethnographic fieldwork (2003: 115).
Despite Temple's and Borchgrevink's calls for language learning and the role of interpreters in social research to be more widely debated within sociology and social anthropology, these issues continue today to attract rather less attention than they deserve. Most surprisingly, perhaps, they are often still not mentioned (or only in passing) in professional codes of ethics, encyclopaedias and prominent textbooks on research methods. It is striking, for example, that there is not a single mention of interpreters -and of the ethical issues that can arise when a researcher works with an interpreter -in the current ethical guidelines of the American Anthropological Association (2012), the American Sociological Association (1997) the Association of Social Anthropologists of the UK and the Commonwealth (2011) or the British Sociological Association (2002). The latest edition of The Routledge Encyclopedia of Social and Cultural Anthropology (Barnard and Spencer, 2012) also does not contain entries on 'interpreters' or 'language learning', nor are these terms to be found in the index of one of the best-known social research methods textbooks (Bryman, 2016). Similarly, the list of questions used when interviewing anthropologists about their field research for a recent book entitled Anthropological Practice: Fieldwork and the Ethnographic Method (Okely, 2012) does not include one on either language learning or working with interpreters, although there are brief references in several chapters to 'learning the language' (2012: 73 and 109. See also Okely, 2007: 66 and 71). The treatment of these issues in O'Reilly's textbook on Ethnographic Methods is only slightly longer, with a paragraph being devoted to each one in turn (O'Reilly, 2012: 95). 5 Although obviously not exhaustive, this brief review of relevant sociological and social anthropological sources nevertheless provides evidence in support of the main point we wish to emphasize here, namely that there remains a dearth of both general discussion and, crucially, practical guidance about language learning and the use of an interpreter in ethnographic research. As far as the second of these is concerned, Borchgrevink (2003: 109-113) has, in a four-page section of the article discussed above, offered ethnographers some very useful concrete advice on how to work with an interpreter. However, as he himself acknowledges, the topic is a complex one (not least because there are different types of interpreter) and he provides only an overview of some of the key issues (2003: 109). Similarly, while the importance of 'learning the language' (if necessary) is usually noted in the literature on ethnographic research, this is rarely, if ever, followed by detailed practical guidance for prospective fieldworkers on how they should actually undertake such a task. In-depth reflexive accounts of how researchers have learned (or failed to learn) new languages for the purposes of fieldwork are also very thin on the ground.
Over the past five years or so, a few notable attempts have nevertheless been made, once again, to draw attention to these issues and in particular to document and analyse in a systematic way the process of language learning in ethnographic research. Alma Gottlieb, for example, has discussed the importance for an anthropologist's professional legitimacy and status of developing 'linguistic expertise', and the challenges of learning a new language for researchers who decide to change fieldsites 'midcareer' or 'in midlife' (2012: 5, 3). Michael Herzfeld (2012) has written a fascinating account of how he learned modern Greek, the Cretan dialect, Italian and Thai and used them in successive fieldwork projects. Annabel Tremlett (2009), for her part, has published a detailed account of her own experience of learning Hungarian before, during and after conducting ethnographic research in a primary school in Hungary as part of her PhD thesis. During her fieldwork, she reflected continuously -in her fieldnotes and in monthly reports she sent to her supervisors -on the process of learning and using Hungarian as the primary language of the research. Tremlett argues persuasively that being aware that she was 'less-than-fluent' in Hungarian led her 'to become more critical and reflexive, to elucidate further and justify the choices made methodologically as well as analytically ' (2009: 80) in the research. In another important contribution, Susan Rodgers (2012) has discussed her experience of learning the Angkola Batak language in the mid-1970s while carrying out fieldwork on the Indonesian island of Sumatra. Based on fieldnotes and the records Rodgers took of her language lessons with several retired schoolteachers, this languagelearning memoir -probably one of the most detailed ever written by an ethnographeralso shows how a researcher can gain many valuable insights into the context in which they are working by a careful and sustained examination of their own language learning in the field. As Rodgers explains, the language-learning education she received from her elderly Angkola Batak teachers also contained 'small, quotidian lessons about language and power' under Suharto's New Order regime (2012: 11).
In the rest of the present article, we draw on this and other recent work, as well as on material from our own research, in order to explore in more detail issues relating to language learning and fluency in ethnographic research. Language learning has in different ways been a crucial aspect of the fieldwork each of us has carried out independently in the past -on the anti-racist movement and then refugee status determination procedures in France (RG) and on nationalism and minorities in and around Moldova and Romania (JDI) -and it is also central to our current joint research project on multilingual working practices in Bulgaria and Romania. However, the doctoral theses and publications we wrote based on our earlier research were effectively silent about our experiences of language learning and other questions arising out of the multilingual nature of the work we had conducted. In what follows, we attempt to rectify this by incorporating examples from our previous and current research into a wider discussion of the importance of language-related issues in ethnographic research.

Language learning and ethnographic research
Although it has long been widely acknowledged that ethnographic research involving participant observation will often require some form of language learning, remarkably little practical advice on the subject exists that has been written specifically for social scientists. Until the 1980s, the general 'textbooks' on language learning most useful to future fieldworkers tended to be those produced by or in association with the Summer Institute of Linguistics or other evangelical/missionary organizations (for example, Nida, 1957;Healey, 1975;Larson, 1984). It was only with the publication of Tonkin's ninepage text 'Language Learning ' (1984) and Robbins Burling's short guide to Learning a Field Language (1984) that relatively detailed guidance, tailored to the particular needs of ethnographers (as opposed to missionaries), became available. Thirty years later, these remain among the very few published general sources offering social scientists practical advice, strategies and techniques for learning a new language in order to carry out their research. For all its merits, however, the approach to language learning Burling outlines is likely to be of limited relevance to many contemporary ethnographers due to the fact that the focus is primarily on the process of learning an 'unwritten' language. The book does contain a useful five-page appendix on 'literary and national languages ' (1984: 107-112), but fieldworkers learning such languages today will need to look elsewhere for more detailed advice. 6 Despite the absence of up-to-date general guides on language learning for social scientists, prospective ethnographers might expect to find some practical tips in accounts written by other researchers of how they themselves learned the language(s) they used in the field. Unfortunately, language learning memoirs are scarce in anthropology, as Rodgers (2012: 11) has noted, and there is little evidence to suggest that they are more common in sociology. Instead, what we often find in the literature are simply statements of the type 'I learned the language', with little if any explanation of how exactly this was accomplished or indication of the level(s) attained in listening, speaking, reading and writing the language(s) in question. A few ethnographers have nevertheless reflected at some length on their own language learning, and we will now discuss more fully what other researchers might learn from the two accounts (Tremlett, 2009;Rodgers, 2012) mentioned in the last section, supplementing these with examples from our own experiences.
Of the many points about language learning highlighted by Tremlett and Rodgers, four are particularly worth noting here. Firstly, the essays by Tremlett and Rodgers show how important it is, on a general level, for ethnographers to record as fully and systematically as possible the processes of language learning in which they are engaged for research purposes. This may seem an obvious point, but it is still worth making, since the collection and subsequent analysis of such material is crucial if the 'mystique' surrounding this aspect of fieldwork is to be dispelled. For example, Rodgers was only able, almost 40 years later, to provide an in-depth account of her language learning experiences in New Order Indonesia because over an 18-month period she had 'recorded the details of all of [her] language lessons with care' (2012: 10). 7 Similarly, Tremlett's decision to reflect on her developing competence in Hungarian in fieldnotes and monthly reports to her PhD supervisors made it possible for her to analyse meticulously not only her own language learning experiences but also the implications of these for different aspects of her research. 8 The second point relates to the amount of time prospective ethnographers estimate initially that it will take them to 'learn the language' sufficiently well to be in a position to conduct fieldwork. Rodgers explains that when she first told her Indonesian language teachers how long she proposed to devote to language learning they 'laughed at my idea that I could master [the Angkola Batak language] in six months ' (2012: 10). She acknowledges that her original assumption was completely unrealistic, but other ethnographers in an equivalent position have probably arrived in the field with similar expectations. What Rodgers describes as 'naive confidence ' (2012: 10), however, can also be viewed, more critically, as an example of the kind of 'intellectual arrogance, cocksureness, or nonchalance' that 'Western' social scientists have been accused of displaying when researching 'non-Western' societies (Owusu, 1978: 327). Be that as it may, the point is that ethnographers must be careful at the outset to avoid seriously underestimating the time they will need to devote to language learning, particularly in a context where funding is often difficult to obtain for lengthy periods of fieldwork. Depending on the time available and the nature of the topic, it may in fact be more sensible for the ethnographer, as Tonkin (1984: 185) advised, to decide instead to work with an interpreter or through a lingua franca (where possible), supporting this with 'informal' language learning and the systematic recording of relevant material (see also , Borchgrevink, 2003: 113;Eriksen, 2004: 56).
Thirdly, an ethnographer who nevertheless does attempt to learn a language for fieldwork purposes can in the process also obtain a different perspective on phenomena relevant to their research. This point comes out particularly clearly in Rodgers' language learning memoir. As Rodgers explains, language lessons and informal conversations over many months with her Angkola Batak teachers provided her not only with 'fieldwork tools' but also 'new eyes to see the New Order [regime]', and more specifically the politics of its 'language dynamics ' (2012: 32, 10). Although she does not use this phrase, her language teachers thus effectively played the additional role of 'key informants'. It is evident that at least one of them acted too as an invaluable source of contacts (2012: 27, 30) and of advice in dealing with the local intelligence police (2012: 24-26).
The relationship Rodgers established with her language teachers is perhaps a little unusual, at least in terms of its intensity and duration, as it apparently involved almost daily interaction over an 18-month period (2012: 10). Nevertheless, what is likely to be true for many other ethnographers as well is that their language teacher is one of the people with whom they have the most frequent contact during their first few weeks or months in the field. An ethnographer's initial impressions of the context in which they are working may well be shaped, in part, by conversations with a language teacher, and the research project itself may even become a topic of discussion in the language lessons, with the teacher spontaneously offering suggestions or advice. This was certainly the experience one of us (RG) had at the start of his fieldwork in Bulgaria for our current joint research project. The teacher he met three or four times a week for intensive language training was one of his main interlocutors for several months, and often commented on aspects of contemporary Bulgarian life relevant to the research. In short, an ethnographer's language teacher may play additional roles (informant, source of contacts, adviser) in the early stages of the fieldwork process; the taking of detailed notes on language lessons can help to bring these to light.
However, and this is the fourth point we wish to emphasize here, language learning in the field obviously continues after or alongside lessons with a teacher in a classroom. Indeed, the kind of immersion in a setting that is often associated with conducting participant observation research can, as Tremlett notes, be 'incredibly enriching for language learning' (2009: 68). Tremlett explains, for example, that when she began to carry out ethnographic research in a Hungarian primary school (after previously attending a series of intensive language courses) she was exposed to a range of new forms of language, including children's slang and jokes (2009: 68-69). In a similar way, one of us (RG) encountered, at the start of his doctoral fieldwork about an anti-racist association in Paris, informal, colloquial and slang registers of French that had not featured in his previous, wholly classroom-based, language education. This quickly led one of the research participants to offer him the following piece of advice: 'You need to learn a bit of French. You speak like Jean-Paul Sartre! (Il faut que tu apprennes un peu de français. Tu parles comme Jean-Paul Sartre!)' In other words, the researcher needed to learn some of the key registers of everyday spoken French, so that he could both understand other people more easily and also communicate with them more naturally, using a less formal or literary register of speech. Herzfeld had a similar experience at the start of his fieldwork in Thailand, when he found that 'the language I encountered on the street seemed very different from what I had formally learned in class ' (2012: 111). As all these examples show, effective language education often involves a combination of language learning in the classroom and 'in the wild', that is, in 'naturalistic' contexts (Pavlenko, 2015). 9 It is important, therefore, as both Rodgers and Tremlett indicate, for ethnographers to continue to reflect on the process of their own language learning in the fieldnotes they take at later stages of their research, even after 'formal' language training has been completed.
In this section, we have been interested in language learning memoirs primarily as sources of information about ethnographers' practices before, during and after fieldwork. However, it is important to remember that such texts are also, in Pavlenko's (2001: 214) words, 'discursive constructions' shaped by social, cultural and historical conventions. Drawing on Kaplan (1994), Pavlenko suggests that language learning memoirs are in fact most appropriately considered as belonging to the literary genre of cross-cultural autobiographical writing (2001: 214-215). Pavlenko's analysis of a corpus of sixteen book-length autobiographies and seven essays (all of which she classifies as 'American') leads her to argue, more specifically, that language learning memoirs are a 'gendered genre ' (2001: 224) in which male and female authors construct their respective narrative voices through different metaphors and other rhetorical strategies and do not attach the same significance to gender as a theme. One of her key findings is that 'male memoirs in the corpus emphasize individual achievements and obscure contributions of others to their language learning', whereas 'female memoirs accord high importance to personal relationships, commitments, and interactions ' (2001: 232, 231). 10 As Pavlenko notes, this difference is consistent with a pattern found in 'Western' autobiographies more generally (2001: 231). Ethnographers writing -and reading -memoirs of language learning for fieldwork purposes need also to be aware of the potential for these accounts to be shaped by such conventions (the precise form of which is of course likely to vary according to the linguistic, historical and socio-cultural context). 11

'Fluency', fieldnotes and forms of identification
Recent work in sociolinguistics has highlighted the ideological nature of notions such as 'a' language, 'bilingual', 'native speaker' and 'mastering' a language (Blanchet, 2016: 51-65) as well as the inadequacy of traditional binary distinctions or classifications (eg between L1 [first language] and L2 [second language] speakers) in second language research (Rampton, 2013: 2, 12). When writing this article, we have struggled to find alternative terms for identifying and exploring language-related issues in ethnographic research. For example, we have found it impossible in the previous sections to avoid using words such as 'fluency' and 'competence' even though, as we discuss further below, they fail to capture the complexity of language learning and use by researchers in the field. While continuing to use such terms here, we nevertheless recognize that one of the challenges for future work in this area will be to develop a set of less reductive concepts with which to analyse the processes concerned.
The lack of attention to language-related issues in ethnographic research is not limited to the main question we have considered thus far, namely how researchers actually learn a second or additional language for fieldwork purposes; it also encompasses the level of 'fluency' or 'competence' in this language they attain in practice, language choice in the context of fieldnote writing, and forms of self and other identification that may occur during the language learning process. As Borchgrevink (2003: 98) has noted, ethnographers rarely write explicitly about their degree of proficiency in the language(s) used in the settings where they conducted their research, still less about the ways their 'language competence' affected their fieldwork. Among those working within social anthropology, he explains, this is no doubt due to a concern that admitting a failure to achieve 'the [disciplinary] ideal of … being fluent in the native language' has the potential to leave a researcher open to 'discrediting charges'; in other words, colleagues might seek to use an acknowledgement of 'lack of complete language mastery' to cast doubt on a fieldworker's authority (2003: 99). It is likely that practitioners of other social science disciplines share a similar fear, at least to some extent, and are in consequence wary of publicly discussing their own language skills in an open and honest manner.
Defining 'fluency' is of course not a straightforward matter, and different research topics may not require the same levels of competence in listening, speaking, reading and writing the languages concerned. 12 Commentators on the matter in relation to ethnographic research often refer to an 'ideal' degree of language proficiency, while simultaneously admitting that few fieldworkers are in fact likely to achieve this in the available time. 13 Burling, for example, advises field researchers that 'the ability to understand native speakers as fluently as they understand each other is the level of proficiency toward which you should aim', but almost in the same breath he concedes that '[t]he majority of field anthropologists probably never reach this level of skill', even after a year in the field (1984: 95, 94). Although Burling encourages those learning a field language not to feel 'dismayed ' (1984: 94) if they fall into this category, it seems reasonable to suppose that a more common reaction is liable to be a sense of failure or at least 'embarrassment' (Tonkin, 1984: 178), and a corresponding reluctance to engage in a public debate about language competence in fieldwork (Borchgrevink, 2003: 101).
One way of helping to remove this obstacle to an open and honest discussion of language issues in ethnographic research would be to acknowledge at the outset that most researchers are likely to be 'less-than-fluent' (Tremlett, 2009: 65) in a new language they have learned for the purpose of conducting fieldwork. Attention could then focus on exploring the implications of this for data collection and analysis, guided in part by the principle that '[e]xplanation of how we conduct our research while still developing field language competence is an important part of revealing our methodology and ourselves as the instruments of data production' (Moore, 2009: 251). This is precisely the approach taken by Tremlett (2009), who reflects in an exemplary manner on her own experience of learning Hungarian before, during and after fieldwork in a Hungarian primary school, and how this affected her choice of research methods in the field, subsequent presentation of the data, and understanding of accountability and 'knowledge claims' in ethnographic research. Tremlett explains, for example, that even after five months in the field, during which time she had conducted participant observation and interviews with teachers and children, she remained concerned about the level of her language skills. This led her to undertake a photography project with some of the primary school children involved in the study, so as to generate additional data 'not wholly dependent on [her] linguistic fluency' (2009: 77). Tremlett's ongoing reflection on her developing language competence in Hungarian also resulted in what she describes as 'a certain fastidious attachment to problematizing and investigating knowledge claims ' (2009: 80). As her language skills improved, she became increasingly aware of the heterogeneous nature of the interactions and other phenomena she was researching, and this in turn reinforced her commitment to an anti-essentialist theoretical position (2009: 75-80). This highlights the way that reflecting on language-related issues in the field can generate insights of importance for ethnographic research more generally.
An important additional implication of being 'less-than-fluent' in a new language is that a researcher may sometimes struggle to follow conversations taking place between two or more other people in a field setting, whilst nevertheless managing without too much difficulty to communicate with individual research participants on a one-to-one basis (including in formal interviews). Lowie (1940: 82) found himself in this situation in 1906 at the start of his fieldwork on the Lemhi (Northern Shoshone) Reservation in Idaho, as, rather more recently, did Tremlett (2009: 69-70) at times during her research in Hungary. There are of course different ways in which a researcher can try to resolve this problem: for example, Lowie ended up working with an interpreter (1940: 83), while Tremlett persevered in Hungarian, albeit with an enhanced awareness of the limits to her knowledge (2009: 70). Regardless of the approach adopted, Borchgrevink's recommendation that '[a researcher] who is not fluent in the language must be particularly diligent in double-checking information and critically testing interpretations ' (2003: 107) is obviously very sound advice. 'Less-than-fluent' researchers who opt not to work with an interpreter need to be especially vigilant in this regard, if they decide to use participant observation as a key method of data collection.
Crucial language-related issues also arise in relation to the writing of fieldnotes in multilingual ethnographic research, and not only for researchers who are developing language skills in the course of fieldwork. How researchers write ethnographic fieldnotes, and the language(s) in which they do so, will vary, depending on individual levels of language competence but also on other factors such as membership of a multilingual research team. What are the advantages and disadvantages of these different individual and collective language practices in the context of fieldnote writing? Detailed answers to this question are extremely difficult to find in the relevant literature. Emerson et al.'s book Writing Ethnographic Fieldnotes (2011), for example, provides practical advice on all the different stages of the process, but, surprisingly, it does not contain any explicit guidance on the basic matter of the language(s) in which to write fieldnotes. The issue is also not discussed by any of the contributors to the recent collection eFieldnotes: The Makings of Anthropology in the Digital World (Sanjek and Tratner, 2016). Among those who have commented (albeit briefly) on the subject is Eriksen, who suggests that 'it may be a good idea to write up one's work in a foreign language (typically English, for nonnative speakers), in order to achieve sufficient distance to local expressions and linguistic categories' in cases where the researcher speaks the language 'too well' and consequently risks taking key aspects of the society for granted (2004: 55, 56).
Most researchers engaged in ethnographic research involving two or more languages, however, probably practise some form of 'translanguaging', moving back and forth between different languages in their fieldnotes. 14 Those who are 'less-than-fluent' in the language(s) used in the field setting nevertheless need to be particularly careful when doing so, as the following example illustrates. During his doctoral fieldwork in France, one of us (RG) wrote his fieldnotes in English (his 'first' language), although he incorporated fragments of French into most sentences. When re-reading these fieldnotes recently, he was able to identify French words or phrases that were obviously unknown to him at the time and which he had noted down incorrectly, for example, 'grenade lacrymose' instead of 'grenade lacrymogène' ('tear-gas canister'), and 'la malle gamme' rather than 'l'amalgame' ('conflation' or 'mixing up'). On some occasions, he had put two question marks in brackets after French words or phrases in the fieldnotes; this was because he had initially noted the latter in the field, subsequently looked them up in a dictionary, but failed to find them (since they had been recorded incorrectly at the outset). In most cases, it would have been easy for a more fluent French speaker (had one been asked) to work out what the correct word or phrase should have been and thereby help the researcher to develop his language competence more quickly and clarify comments in his fieldnotes. This highlights once again how important it is for 'less-than-fluent' fieldworkers to devise a range of additional strategies to ensure the accuracy and reliability of the data they collect, while of course protecting the anonymity of research participants.
In the final part of this section, we wish to draw attention briefly to one further language-related issue arising in multilingual ethnographic research that has also been rather neglected by social scientists and which deserves to be discussed more widely. As many studies of second language education have shown, the process of language learning can be affected by a range of social factors, including unequal access to language classes, classroom interactions, the power relationship between different languages in the world, and socio-political changes (see, for example, Pavlenko and Piller, 2008;Lightbown and Spada, 2013: 89-90, 150-151). This is obviously as true for fieldworkers learning a second or additional language for research purposes as it is for learners motivated by other reasons, and the former therefore need to document and analyse such influences carefully too. Recent sociological research on language and 'identity' points to one way in which ethnographers can start to explore these questions. Drawing on findings from a study of the 'narrative identity' of Polish speakers in the North of England, Temple has argued, for example, that 'decisions about learning languages are influenced by wider concerns of self and other identification rather than simply being issues of instrumental need ' (2010: 287). More specifically, she shows how research participants' views about learning English were influenced, among other things, by the importance of speaking Polish for their sense of 'self' and the way in which language featured in distinctions they drew between 'Polish' and 'English' values (2010: 289-293). 15 How ethnographers approach and experience the process of language learning in fieldwork is also likely to be shaped, to some extent at least, by forms of self and other identification.
An illustration of this last point is provided by the importance one of us (JDI) attached, at the start of fieldwork in Romania for our current joint project, to being once again able to 'pass for a native Romanian speaker', both socially and linguistically. Ten years previously he had spent two years in Bucharest, as first an Erasmus and then a Master's student, and it was during this time that he developed the desire to speak Romanian like a 'native'. He would analyse the latter now in terms of self and other identification in the context of the relationships he formed with other foreigners and Romanians. On the one hand, his aim to 'pass for a native' was a way of distancing himself from the category of 'tourist language learner' (Phipps, 2007) and from what he regarded as the 'bubble' in which many other Erasmus students lived, as well as a reaction against the assumption many people made that as a French speaker he would not make the effort of learning to speak Romanian well; on the other, it was underpinned by a concern with 'being a member' (Davies, 2003: 99) and with showing respect to the Romanian people he met. 16 Other ethnographers who have attained a high level of fluency in a second or additional language are likely to have had similar preoccupations at some stage in their language learning.

Conclusion
Conducting ethnographic research often involves 'researching multilingually' (Holmes et al., 2013), that is, using two or more languages during the different stages of the research process. Building on the work of scholars such as Temple (1997), Borchgrevink (2003), Tremlett (2009) and Rodgers (2012), we have sought in this article to dispel a little more of the 'silence' or 'mystique' that, in our view, continues to surround two key language-related issues in ethnographic research: language learning for fieldwork purposes and the possible effects on research practice of different levels of fluency in a second or additional language. On the first of these themes, our central contention has been that a comprehensive, 'demystified' account of the process of conducting multilingual ethnographic research must include the detailed description and analysis of any language learning that took place during fieldwork (and, ideally, also an examination of any relevant language study the researcher undertook before and after being in the field). 17 We hope that other ethnographers, working in different parts of the world and in different languages, will be encouraged to publish their own language learning memoirs, not only to aid future fieldworkers embarking on the process but also so that a sizeable corpus of such texts becomes available for comparative analysis.
On the second theme, we have argued in favour of explicitly acknowledging that many ethnographers who set out to learn a new language in order to carry out research will probably remain, despite their best efforts, 'less-than-fluent' in it over the course of their fieldwork. Recognizing this could facilitate the development of a much-needed discussion of the implications of different levels of fluency for data collection and analysis. We have sought to contribute to the understanding of such issues by examining the potential impact of degrees of language competence on choice of research methods, the writing of ethnographic fieldnotes, and forms of self and other identification on the part of the ethnographer. The latter are important matters for all ethnographers to consider, and reflecting on language learning and use can help to develop understanding of them. The analysis presented here could usefully be extended to include how levels of fluency affect other aspects of the research process, not least the translation and interpretation of fieldwork data. Wider discussion of all these matters in ethnographic research would help researchers make more informed choices when carrying out and writing up research using different languages, enabling them to act more 'purposefully', in the sense of 'being able to articulate the rationale for their researching multilingually choices, rather than simply stating what they did' (Holmes et al., 2013: 297). In so doing, researchers in sociology and social anthropology could learn from as well as contribute to fields such as linguistic ethnography (Copland and Creese, 2015) that investigate language use in relation to wider social processes. Ethnographers have much to gain by breaking the silence (again) about language-related issues in their work.