“If you’ve done a good job, it’s as if you’ve never existed”: Translators on translation in development projects in the Sahel

Translation is an essential and extensively-used tool in research and development projects, yet is frequently sidelined as an insigni ﬁ cant or minor component in the initial design. This often leads to assumptions regarding translation tasks, by both the translator and the end-user or the commissioner. Addressing this lack of awareness and the resultant misunderstandings concerning the translation outcome, this article examines translation processes that take place when translations are commissioned. It draws on empirical data from an NGO radio development project in Africa ’ s Sahel, including semi-structured interviews with translators working from Fulfuldé, Tamashek, and Zarma-Songhai into French with responses clustered around four themes: identity, agency, source text knowledge and transcription/translation processes. Contributing to translation studies and to development studies, the article provides recommendations on implementing changes to overcome dismissive attitudes towards translation, and to promote its consideration as a core element of development and research projects.


Introduction
Translation is an essential and extensively-used tool in cross-cultural research and development work, yet is frequently sidelined as being of little importance and only a minor component of the initial project design. Assessing the role of translation in field studies, this article draws on a larger project, femmepowermentAfrique, which assesses radio's impact on women's rights and empowerment in Niger, Mali and Burkina Faso. 1 This impact assessment was requested by two non-profit media development agencies: the Swiss-based Fondation Hirondelle and the Danish-based International Media Support (IMS), both seeking to provide independent information to communities and particularly women in conflict-and crisis-affected regions. Data for the impact assessment comes from a series of focus groups, knowledge-exchange workshops, semi-structured interviews and detailed content analysis of sample radio output. Using Nvivo, this analysis examines the positivity and negativity of individual terms and phrases, the use of stereotypes, geographical references, the proportion of airtime taken by male and female voices, dominant and neglected themes, and so on.
Although Fondation Hirondelle and IMS have similar goals, they use different broadcasting models. Fondation Hirondelle operates studios in capital cities (in this case, Studio Kalangou in Niamey, Niger and Studio Tamani in Bamako, Mali) that broadcast daily news and discussion programmes. These programmes are transmitted via satellite to partner (commercial or community) radios located throughout the country, who then re-broadcast them locally using FM networks. The studios call on local correspondents as a means of mitigating this "top-down" model. In a contrasting model, rather than producing programmes from each capital and relaying them to the regions, IMS'sSahelProgramme (2018)(2019)(2020)(2021), operating in the conflict-ridden border area between Mali, Burkina Faso and Niger (IMS n.d.), works with women's clubs d'écoute [listening groups] associated with local radio stations to produce locally-generated content. Because these clubs d'écoute include people who are familiar with local issues, they can determine topics of particular regional interest and relevance and work with local radio stations to produce programmes germane to their audiences. IMS also provides content and journalistic training to both the clubs d'écoute and the radio stations, in order to reinforce local community radio provision and strengthen women's voices in local radio and society.
All the material collected from Fondation Hirondelle is in Frenchan official language in all three countries, and, as a result of the colonial past, the language of government, business, media and educationand is being transcribed and analysed as French texts. IMS's material, however, is in three different languages that hold varying status as official or national languages. Tamashek is the language of the Tuaregs, spoken mainly in Mali but also increasingly in Niger and Burkina Faso. 2 Fulfuldé is a large dialect continuum with many alternative names such as Fula, Fulani, Peul, Pular, and is widely spoken across the Sahel and West Africa by the Fulbe people (Breedveld 2009). Zarma is one of the Songhai languages and is the most widely spoken indigenous language in Niger (Bornand 2006;Hamani 1982).
Given femmepowermentAfrique's lack of working knowledge of these languages, the IMS content first needed to be translated into French, the project's working language, before being analysed. Commissioned and carried out by five local translators, it is these processes of translationfrom oral to written text and then from Zarma-Songhai, Fulfuldé and Tamashek to Frenchthat are the focus of this article in its examination of the role of translation in a development project. Drawing on empiricallygrounded data, the article contributes to translation studies by widening the scope of translation practices and the pool of languages being researched. It also provides specific, real-life examples of these practices and languages in regions which remain under-researched and under-represented in the field. The article also contributes to development studies by providing practical recommendations for implementing changes in order to overcome dismissive and neglectful attitudes towards translation and to promote its consideration as a core element of development projects.
We situate the project with regard to women's empowerment, radio and multilingualism in the Sahel region, then discuss the mutual importance of translation to development studies and development to translation studies. Following a methodological overview in which we present our three research questions, we analyse and discuss our data around our four interview themes: the translators' perceptions of (1) their role and identity; (2) the source text; (3) their own agency, and (4) the processes they use to complete the task. The findings are then discussed in terms of recommendations for use by those working on future cross-cultural research and development projects.
Women's empowerment, radio and multilingualism in the Sahel region Measuring the impact of any activities which target women's empowerment is essential, especially in the Sahel where high gender inequality prevails. In all three countries examined here, women are subject to high child marriage rates (UNFPA 2012), low schoolcompletion rates and high illiteracy levels (Save the Children 2016), and polygamy is widespread. There is extensive violence against women, female genital mutilation (FGM) is common (Thomson Reuters 2018) and legally, women do not have the same status as men. Despite displaying extensive agency within some sectors of society, women remain disempowered and require more information about their rights to achieve a greater voice in society (Heywood and Tomlinson 2019).
Radio remains one of the main sources of information available to women as a marginalized group. As a secondary medium (Berland 1990, 179;Chignell 2009, 99;Fleming 2002, 1), radio reaches women engaged in domestic chores to which they are typically committed in patriarchal and traditional societies and is cheap and easily accessible via mobile phone (Heywood 2018). Radio also has the socio-cultural advantage of being able to reach illiterate and orally-based cultures where multiple vernacular languages are spoken. This is particularly relevant in many African countries and is an invaluable tool in promoting the power of oral traditions, which prevail amongst many communities (Girard 2007;Jallov 2011).
National languages abound and many are recognized officially. Niger, for example, has eleven official languages, Mali has thirteen and Burkina Faso has eight (Ousmane Ida 2015; Panis 2017). There are many more local languages (up to 70 in Burkina Faso), and many are ethnically and linguistically closely related. 3 As Skattum states, determining the exact numbers of local languages in the region is difficult because of the lack of research, migrations, dialects, language shift and multilingualism (2008,104). The situation is rendered more complex by the fact that an individual's language of primary use, for example in their workplace, often differs from their mother tongue. Formal education is widely conducted in French, but as only a small percentage of the population formally attend school, a correspondingly small number of people speak French despite it being the official language (13 per cent in Niger, 17 per cent in Mali and 24 per cent in Burkina Faso [OIF 2019]). Multilingualism is key with many local languages being understood by different ethnicities, something which emerged in IMS's radio broadcasts. Whilst linguistic plurality can be perceived as an active and useful strategy for participating in social life (McLaughlin 2015), and although boundaries between ethnic groups in the region are uncertain due to the mixing and assimilation of populations, language is also representative of groups and families and can act as indications of the dominance of certain ethnic groups (Skattum 2008).

Translation and translators in development studies
In development studies, translationoften understood as encompassing various forms and combinations of written translation, oral interpreting and intercultural mediationis often viewed as an administrative task occurring after the generation of texts and allocated minimal time and resources in grant applications and research budgets. This was confirmed by The Listening Zones of NGOs, the first major investigation into the role of languages, cultural knowledge and translation in development programmes. The project found that "languages have a generally low priority in development work"; that "languages are not generally integrated into the development cycle and budgeted for in advance"; that "the potential need for translation within development projects is often an afterthought"; and that "discussions on sustainability/supporting local capacity rarely include building on and nurturing the linguistic potential of communities" (Footitt, Crack, and Tesseur 2018,4 -7). This is in spite of additional key findings that "paying close attention to languages is seen as contributing to successful development"; that "failure to integrate languages into development initiatives leads to certain groups being excluded"; and that "the mutual 'untranslatability' of concepts" was commonly encountered (Footitt, Crack, and Tesseur 2018, 5).
Evidence shows increasing awareness from practitioners and academics of the importance of translation to cross-cultural research (e.g. Floro and Komatsu 2011;Maclean 2007;Temple and Edwards 2002), development and humanitarian aid (Cornwall and Eade 2010;Crack 2019;Federici et al. 2019;Footitt 2017;Powell 2006;Tesseur 2017;. There is also a growing body of literature concerning cultural translation and "vernacularization" of concepts across "local" and "global" contexts Merry 2009, 2011;Østebø 2015;Unnithan and Heitmeyer 2014). The Listening Zones project reports on innovative practice from the field (Footitt, Crack, and Tesseur 2018,2 8 -31), including collaborative and participatory translation workshops; the development of communication plans, glossaries 4 and handbooks that consider local languages and translation needs; and the use of back translation to ensure consistency across languages. Nevertheless, empirical and theoretical investigations into the translation processes necessary for the planning, delivery, assessment, evaluation and reporting of development projects remain limited. Similarly, the identity, daily work and agency of the translators themselves also remain under-investigated (although see Delgado Luchner and Kherbiche 2018;Wright 2018).
When discussed in the development studies literature, translation is frequently used metaphorically. Translation as negotiation between actors (Lewis and Mosse 2006), the translation of research into policy, of policy into practice, the differing interpretations of goals and ideas through an organization'sd i fferent layers, the translation of (often "Western") values into new contexts and value systems, and so on, are common themes, often expressed in terms of an undesirable "lost in translation" scenario. While broader, figurative conceptualisations of translation are increasingly accepted and explored in translation studies, their use in development studies risks overlooking interlingual and intralingual translation as integral parts of those broadly defined translations and interpretations.
This broader, metaphorical notion of translation is, however, relevant to our case study where we (and Fondation Hirondelle and IMS) are interested in how impact studies feed back both to radio content providers and to project managers and donors. There are assumptions here that translating the focus groups results into evolving radio policies and practices will, in turn, translate into social and political changes on the ground. That intralingual translation (from oral recordings into written transcripts) and interlingual translation (from Fulfuldé, Tamashek and Zarma-Songhai into French; and from French into Fulfuldé, Tamashek and Zarma-Songhai) are essential elements necessary for these larger, metaphorical translations of change to occur seems self-evident. Yet these translation processes and communicative contexts, involving real resources and real people, including transcriber-translators, are largely assumed in the commissioning of the impact study and consequently, in the design and delivery of the impact study. Even when the issue of interlingual translation is problematized -IMS commented that the necessary translation would be "hard" and the process "complicated"-and budgets are allocated for translation, the practicalities of finding (competent? trained? appropriate? available? reliable? enough?) translators and negotiating fair rates, usually through a middleman, can be, as they were in our case, challenging. Additionally, the importance of a clear and detailed brief, where the expectations of NGO(s), researcher(s), broker(s) and translator(s) are successfully communicated up and down the supply chain, so to speak, becomes crucial.

Development and translation studies
Just as the role of translation in development and cross-cultural research is often invisible, ad hoc, unaccounted for and under-theorized, so too has little attention been paid to this by scholars working in translation studies. Even as the field becomes increasingly diverse and internationalized, it remains dominated by a relatively limited pool of language combinations and theoretical perspectives. Investigating the translation practices taking place in this one example of a development project enables us to focus on the often-overlooked role of translation in development, and also consider (for the first time, as far as we are aware) translation taking place between languages, by people, in places and for purposes starkly underrepresented in translation studies.
Apart from the recent work by the Listening Zones project, Kobus Marais' monograph (2014) is the most sustained attempt so far to unite translation and development studies as a means of addressing the blind spots around development and interlingual and intercultural communication common to both disciplines. Marais' aim "is setting an agenda or a framework for future work" (2014, 120), and our case study responds to his call "that translation scholars … work empirically to trace and account for the interfaces of translational and developmental actions" (144). Our study aligns with Marais' call for considering the informal economy (ILO 2018) and alternative and diverse modes of agency that differ from more traditional notions of political activism. These contexts, we agree, have "not been in the purview of translation studies scholars", in spite of their pervasiveness globally, thus not only skewing "the picture we have of translation" but skewing that picture "toward developed countries" (2014,145). To widen the agenda, Marais asks: Do we know the role of translation in development? Do we know which texts are being translated? Do we know how people communicate in development projects where various cultures, values, and languages meet? Do we know what happens in oral translation in development? Do we know what constraints developing contexts place on translators and their translations? And, can we know these things and how can we know them? (2014,145) Marais seeks to answer these questions. He combines textual analysis and empirical data from interviews and ethnographic observations in several South African locations to examine the (non)translation of local economic development policy documents, the translation of academic agricultural knowledge to farmers and local stakeholders and translation in the multilingual informal economy. Our article adds to this. By critically interrogating the translation processes intrinsic to the impact study of "development projects where various cultures, values, and languages meet" (Marais 2014, 145), we offer empirically-based evidence regarding which texts are being translated and the constraints developing contexts place on translators and their translations. We reflect on the limitations of our study and methodology, thus addressing Marais' epistemological questions.

Methodology
Hour-long, semi-structured interviews were conducted in Niamey in January 2019 with each of the five transcribers/translators working on audio recordings in Fulfuldé, Tamashek and Zarma-Songhai into French: Adamou, Ali, Bouba, Issouf and Moussa. 5 We initially assumed that the transcribers of the recordings would be different people from those who then translated the transcriptions from local languages into French, and so we developed parallel interview questions for transcribers and translators. The focus on transcription was deliberate. As a form of intersemiotic translation (from oral to written mode) widely-used in development projects, cross-cultural research and oral history and yet little reflected upon (Temple 2013), we sought to analyse this task as much as the subsequent interlingual translation. The transcribers and translators were, in fact, the same people and therefore the questions were asked only once and, in their answers, respondents often conflated references to both transcriptions and translations, or emphasised one over the other.
All the interviewees had, at the time of the interviews, transcribed and translated at least one broadcast. Semi-structured interviews provide a "conversation with a purpose" yet ensure some spontaneity as interviewers can add follow-up questions in response to answers (Galletta 2012;Gillham 2000;Kvale 2007). The interviews were conducted in French by Heywood, using the same questions and similar probes as detailed below. The interviews were audio-recorded, transcribed and coded using thematic analysis. Participation in the study was voluntary but respondents' expenses were covered. The quotations here are Heywood's translations of their answers.
Although semi-structured interviews may be perceived as an easy data collection method (Wengraf 2001), they become complex when the many social layers at play in this context are considered. For example, like the interviewer, none of the respondents claimed French as their mother tongue, a clear example of French being used as a langue de communication as Ali, one of the interviewees, called it. Additionally, all the participants were maleunsurprising in the Sahel's male-dominated society, especially given labour inequalitiesand were appointed by the project's fixer, also a man who was possibly unaware of the gender imbalance which resulted from his choices. Studio Kalangou does, in fact, employ female translators showing that not only do the latter exist in Niger but that they could have been sourced for the larger project.
Social desirability bias, or the "tendency to say things which place the speaker in a favourable light" (Nederhof 1985, 264), must also be considered. Respondents may give answers to avoid negative evaluation or to gain endorsement or approval from the interviewer. This is relevant here as the respondents were being questioned about their work practices by the person who was going to use that very work. Avoiding such bias in this context could only be aspirational. Therefore, the interviewer made all attempts to place respondents at ease, refrain from judgemental comments, and allow respondents to express themselves freely and at length.
The research questions are: . RQ1 How do translators/transcribers working for NGOs in an informal economy perceive their own work, their own role, and the methods they used?
. RQ2 To what extent do these perceptions align with the expectations of the end-user as expressed in the instructions?
. RQ3 How can differing expectations be overcome?
To answer the research questions, the interviews addressed four themes: translator identity, translators' knowledge of the source texts, translator agency and transcription and translation processes at both textual and extratextual levels. The first question (Tell us about your work. What do you do?) was open-ended, inviting translators to talk about their work on their own terms. Motivated by debates in translation studies around the agency of translators which tend to emphasize the causal effect of the individual translator, usually in terms of their politics, ideology or narrative position (Marais 2014, 89), the question sought to ascertain how each translator described themselves and their work and spoke about their own agency. We asked about the respondents' langue maternelle and their knowledge of, and attitudes towards, other local languages and French. This was to elicit some reflection on identity and power differentials given the varying status between local languages and the colonial language. Insight from translators into the multilingual contexts in which they work (as described above) was seen as valuable, given our own limited perceptions of these contexts, and also given the roles, status and use of these languages in the work of Fondation Hirondelle and IMS, where top-down approaches often assume and rely on translation from French into local languages, and bottom-up approaches assume and rely on translation from local languages into French.
The second group of questions concerned interviewees' knowledge about and perspectives on the (source) texts they were transcribing and translating. This drew on translation theory's emphasis on source-text analysis. Championed by functionalist approaches, comprehensive and systematic source-text analysis is seen by many as the first step in any translation task, "since this appears to be the only way of ensuring that the source text has been wholly and correctly understood" (Nord 2005, 1). Furthermore, functionalists argue, source-text analysis "should also provide a reliable foundation for each and every decision which the translator has to make in a particular translation process" (ibid). While we did not expect such analysis to be undertaken as a separate taskit normally becomes an integral, even unconscious, part of the experienced translator's expertisewe were interested in the knowledge our translators had (been given) about the source texts and the extent to which this knowledge informed their practice, especially when the audiences (radio listeners, researchers) and purposes (to inform, to evaluate) between original and translations differ.
The third group of questions returned to the issue of agency and invited interviewees to reflect on their work in terms of wider social and political contexts. We examined their role as translators, whether they were proud of their work, what they liked and disliked about translating these texts, and whether they thought being male impacted their work. We also asked whether they were proud of being a translator, as we were interested in how this position was regarded by others in their society.
The final set of questions concerned processes. We asked about the instructions given; the practical and material considerations of their work; their cognitive processes and problem-solving strategies; self-evaluation and feedback. The aim was to gather empirical evidence on the nature of translation practices in the local informal economy, both to expand the purview of translation studies, which has largely been concerned with "the elite component of societies worldwide" (Marais 2014, 194) and to consider the practicalities of translation in development, which are commonly assumed, under-interrogated and undocumented.

Analysis and discussion
Question Set 1: Translators' role and identity Two of the translators described their training and work experience in some detail. Ali said he had "supervised the experimental Fulfuldé bilingual schools" and worked "for the publishers of school textbooks" as well as doing some freelance translation for individuals. Bouba, the author of several books in Zarma, with a degree in journalism and himself a product of the experimental schools, acknowledged that languages have benefitted him, allowing him to travel and secure good jobs in the media and elsewhere. Much of his work has been with NGOs, with translation/interpreting (he does not differentiate) a key component of that work. "Ip a s so nt h em e s s a g e ", he says repeatedly of his translation work. "You have to find the right word to get the message across … There was one project on how the right message should be passed from person A to person B. I translated it all". Bouba was the only respondent to talk about translation with this sense of purpose, to inform and communicate: "You need to ask not what they're being told," he said, "but what they understand, to bring about change". For the remaining respondents, translation was described in purely mechanical termsfor example, "I do translations and transcriptions" (Moussa) -referring to contracts, language directionality, the need for concentration, and the amount of time it takes. Nobody mentioned any training in transcription or translation.
Nevertheless, subsequent questioning about the translation's textual processes revealed that the respondents clearly differentiated between transcription and translation work. Transcription demands exactness, whereas translation requires different strategies and decision-making. They said that transcription "means reproducing exactly what the person being interviewed has said" (Ali); "No summaries … I write word for word what's said and remain true to that" (Moussa); "I write down what I hear" (Adamou); "Every word that's said on the audio, I listened and transcribed. Even repetitions" (Issouf). In contrast, with translation they felt able, even compelled, to make changes to the text: "Word for word translations betray the original" (Bouba); Adamou "often adds things" for clarity; Moussa summarizes because "In Tamashek, phrases are kilometric … but in French, I can't put in these repetitions. I summarize [and] put it in a single phrase". He also "explains what the audio is saying", as his role is to "get across the message". Yetthisapparentdichotomybetweentranscription and translation is not clear-cut and the translators discussed how they made decisions about the text during transcription. Ali said he added clarifications in brackets if speakers "swallow their words";A d a m o us ai dh eo m i t t e d repetitions or what he considers superfluities: "Some words don't have any meaning in the local language so I ignore them. If they don'tm e a na n y t h i n gt h e y 're not included".W h e n asked whether they indicated pauses, interruptions, laughter or other sounds, only Issouf said that he indicated laughter, with "hahahaha" or exclamation marks; whereas Moussa said "Ic a n 'ts h o wt h e s e .H o wd oIw r i t et h i sd o w n[ h ec l a p sh a n d s ] ?Id o n 'ti n d i c a t e laughs. I just transcribe the words". The gender of the speakers was sometimes indicated -"the names show this" (Ali); "Ip u tparticipantet os h o wi t 's a woman" (Bouba) -but not always, and the multilingual, multi-dialect nature of the listening clubs was also flattened or removed from the translations. "She speaks in Zarma",M ou ssasa i do fon evoi c eont h e recordings. "She understands Tamashek but doesn't speak it. Everyone understands what she'ssayingtoo .… I transcribed her words in Zarma, then translated it into French".
These rudimentary theorizations of transcription and translation illustrate how both processes are more complicated and contradictory than commonly assumed. When we asked about the instructions the translators were given, it seemed they received very little: "I wasn't given any instructions. Listen, transcribe, translate. That's all" (Ali); "The instructions were simply to do the transcription and translation. It was up to me to determine the methodology" (Moussa); "I was actually told to put down the contents, not lose the meaning, listen to the audio and write down what was said, and not modify the contents. It's the contents which are important" (Issouf). These responses also suggest that it is further assumed that everybody involved, end-users, middleman and translators alike, hold shared understandings of exactly what is meant by, and required of, transcription and translation.
Social desirability bias may also be at play. As noted below, feedback on work is rare and these interviews appeared to be the first time that these translators had discussed their work with anyone in detail. Thus, their insistence on transcribing everything they heard and on conveying the "message" and the "contents" in their translations emerged as somewhat expected defences of their work, with the above assumptions forming the basis for such defences and the lack of instructions acting as a disclaimer, protecting them against possible criticism. That these insistences on accuracy and message transference begin to break down, when asked details about the actual textual processes and decisions, indicates the inadequacies of these assumptions and instructions. Addressing these inadequacies form a major part of our recommendations and is discussed below.
The role and agency of the translator was pursued in Question Set 3 (below), but already the prominent role of language in how the translators spoke about themselves and their work emerged strongly in these opening responses. This was clearly, even passionately, expressed when we asked the translators to talk about their native language and its meaning for them. For Ali, Fulfuldé was his "raison d'être": Fulfuldé has to live because Fulfuldé … is a tradition. If everyone understood the tradition and customs and behaved [accordingly] they wouldn't be any wars. No one would think they are better than others. That's why I want Fulfuldé to live and exist.
Adamou said "I'm Songhai. I use it for research, for work, as a tool but it's more special to me than that. Ma langue est mon tout. [My language is my everything]". Issouf said "You've seen the tribes of Zarma people who dance? Well, that's us. Real Songhai. I'ma true Songhai".
In contrast, other languages considered useful by the five multilingual translators were used "for work, or with friends who don't speak Tamashek" (Moussa), "just to communicate" (Ali), or out of necessity: "I learnt Hausa because I had to …"(Issouf). French was described as une langue de culture, de communication and the language of education, and all spoke extensively about the grammatical, syntactic and lexical differences between French and their language that pose challenges to them as translators. Identity and any sense of agency, in terms of individual politics or social motivation, are linked to linguistic rather than translation expertise.

Question Set 2: Source texts
All the translators grasped the significance of gender permeating the texts. They were also aware that these were community radio broadcasts with a broad audience. Two of the translators understood our question about the purpose of the source text ("Who are the texts for?") to be about the purpose of the target text. For Issouf, this was important -"I understand it's a study being done by researchers and I've really taken my time to correctly transcribe the information"-whereas for Moussa it was not: "I don't have any details about that. I was simply contacted a few days earlier, they said I've got some work for you, I need someone who speaks Tamashek and can translate into French".
The subject matter of the source text was then approachedwhether it was difficult or distressing, whether it personally or emotionally affected the translatorsand was motivated by emerging debates in translation studies around the effects on translators or interpreters when translating traumatic material (Elias-Bursać and Askew 2018). Our interest was the extent to which translating material related to ongoing violent conflictaffected translators who are themselves members of the communities subject to the violence.
Any effect seemed to be minimal. Answers included indifference -" it's just a job" (Issouf); "I've had to translate things which don't really interest me … I'm paid to do a translation so I do a translation" (Ali) -and professionalism: "I do a lot on conflict, cutting [FGM], deaths. … It affects me but I know that I've a proper job to do to help others" (Bouba), which Moussa expressed in terms of loyalty to the commissioner and the source text: "Obviously, it affects me. I'm human. Whatever the subject, I still translate it. I can't betray the person who has given me the work. But it can still shock me. But I don't change things. If there are swear words, I would translate them".
Interestingly, if issues about the subject matter of the source texts were raised, they were in terms of social sensitivities rather than personal distress. Whilst transcription was considered inviolable, it was noted that the original broadcasts were themselves sensitized: "Because it's on the radio they have to be careful and respect things. The words they used for rape have already been carefully chosen. … It's a question of culture. They've already screened the work" (Bouba). The transcription could not change anything: "You have to put down the contents as they are … If it's a sensitive subject, you can't cover up the issue" (Issouf). Two translators, however, talked about the need to adjust to local sensitivities. For Bouba, working as a "conflict journalist", translation can be used to mitigate a language's violence in a way that is sensitive to a region suffering from violent conflict, even as he recognizes the importance of using powerful language. He says, "We were taught there are things you mustn't do. 'He was killed, had his throat cut' you shouldn't say these things. There's a way of saying them. … In translations, the languages vary. I prefer to use words which are less shocking".
Adamou described how issues of sensitivity become even more acute when interpreting.
If it's translation and on paper, it's ok, even if it's a taboo subject. You don't have the same pressure as when you have to say it in public. That's when the problems arise. Once I was translating in public and someone asked about male or female sexual organs and, really, saying that in public … To start with, we used more general words but specialists told us to be clear and use specific words.
What is interesting here is that the translator feels required to respect the perceived sensitivities of local audiences rather than any that may be perceived to be held by the NGOs.
As members themselves of that community, the conflict between local and NGO expectations is felt by interpreters who are physically present during the exchange, and who find it difficult to take on the roles afforded (relatively) anonymous written translators of mitigating sensitivities or not changing anything, as described above.

Question Set 3: Translator agency
The third set of questions returned to agency. Pride in their work and their role was expressed in terms of thoroughness and professionalism: "Yes [I am proud of my work] because I know I've worked hard" (Issouf); "Very proud. Having corrected it and made sure it's all in order, that it'so k " (Moussa). Ali, who had earlier spoken about his pride in Fulfuldé, expressed pride in his translation work through reference to his language: "I'm proud of my Fulfuldé work, getting it out there so the world can understand, so that people can understand that Fulfuldé is a rich and subtle language and that they need to learn it". None thought that being male had any effect on their work as translator. "No impact", said Moussa, "Even if a woman did understand giving birth better than me, they still have to translate what is written down. We're here to translate not to give our opinions". Ali and Issouf's responses concurred, yet both spoke about the importance of the work contributing positively towards social change: "I know some subjects are taboo in society but unless we talk about them they will carry on", Ali said. "People need to understand and tackle taboos. But the difficulty is getting subjects like these to be discussed in the home". Ioussef said, … I prepare myself psychologically because there are things which are shocking. I run workshop on sensitive subjects, for example how to manage pregnancy, birth … People confuse religion and culture, so they think something is taboo. But if you explain to them, they understand. Say a young girl of 15 or 16, if you tell her that if she sleeps with a man she'll get pregnant, she'll know. But if you don't tell her, or don't tell her about precautions, how's she going to know? How can she protect herself? People think it's taboo but then only discuss it when they're caught. They say it's God's will, but that's utopique.
Issouf's unconscious biashe began by stating that being male has no impact on his work emerges in his account, which is given completely from a man's viewpoint. He fails to comment on men's role in this scenario; the onus is on the girl to be informed, to know about precautions, and to protect herself. Translators' embeddedness in the society in which they work is largely unreflected upon, as we saw above with the insistence on remaining unaffected by the contents of the broadcasts.
Yet, there was also evidence that this is not always the case, and three translators commented on positive aspects of the work. I now feel that women can serve society, the community as much as men. That's why I'm loving it; hearing about rural areas and this shift … I learn things because before listening to the audio. It's shaped me too. (Ali) I've no problem with the topic. It's interesting especially regarding women and empowerment. (Adamou) I like themes which affect me. For example, one woman said, ' never knew that the boats transporting migrants had women in them'. I learnt from the broadcasts. (Bouba) The work of the radio is potentially also reaching the translators and their families and networks.

Question Set 4: Transcription and translation processes
The fourth set of questions focused on the physical environments in which the work occurred and on the networks of people and objects that the translators created for their work. The aim was to empirically add to the picture (and theories) we have of translation and to consider processes and networks that are largely invisible or assumed whenever translation is commissioned by NGOs.
All the translators used computers -Ali mentioned "software for the local languages for their specific letters"-although Adamou uses pen and paper, because, as he explained, "I'm not quick on computers". He, therefore, transcribes and translates by hand and then works with someone else "to type up. So we work as a pair". Issouf states he has "two windows open on my computer, the audio and the transcription, and I flick between them", but still "write[s] on paper then type[s] it up. Otherwise, if you go straight to typing up, you might make mistakes that change the meaning". This handwritten work adds considerably to the amount of time required; the translators mentioned redrafting and re-reading, and checking and transcription were described as especially time-consuming; Adamou described how he listens to the audio several times before starting: "This work is not easy, it's really boring. Because you have to keep stopping and starting. It takes a huge amount of time". All the translators talked about long hours, the need for concentration, working at home, in the evenings, at night and at weekends: "the work is infernal", Ali said. "The subject is ok but the work is stressful".
Nobody mentioned assistive tools such as voice-recognition or transcription software, standard for most freelancers in the developed world. The lack of dictionaries was a concern; some lexicons were available in Zarma-Songhai, none in Tamashek, and Ali talked about building his own Fulfuldé lexique, especially for borrowed words, on which he then consults with a local group of colleagues which works on local languages.
This practice of collaboration appeared common. In addition to the example where Adamou works with a typist, the translators talked about consulting colleagues; "I check with others about certain words. There are regional differences … especially when it's scientific. I turn to colleagues and ask how they would define something in the local language. We're used to working together" (Moussa). Such collaborative work differs considerably from the way translation is normally theorized as a very individual practice, and has repercussions on issues such as time, payment and confidentiality, which have yet to be considered.
In contrast to this collaboration, the feedback the translators received seemed to be minimal, although contradictions emerged, most likely because of the broadness of our question. Feedback seemed to assume some kind of criticism or disapproval. "I don't get comments. I haven't encountered any difficulties" (Moussa); "I don't get feedbackneither good or bad. I therefore assume it's good" (Adamou); "If you don't respect the instructions then there's feedback and that can break the trust between you and the client" (Issouf).
Translators are similarly reticent about directing questions to the commissioner. "I don't ask any questions, just the person who asked me to do the work" (Moussa); "Even if they do ask questions, it's rarely from any kind of negotiating position of expertise. In this case, when I started I asked if I had to transcribe it word for word or my way, and he said it would be better to write down what I hear, and that's what I did" (Adamou). Without any systematic or constructive feedback, responsibility for, and pride in, the work remains a very individual, idiosyncratic affair. "I know it's ok as I have respected the original and done it word for word. I haven't added anything or left anything out" (Moussa).

Conclusions and recommendations
This article discusses real-life examples and experiences with translators on a project in Niger. Whilst it focuses on transcription and interlingual translation processes integral to the impact assessment of a larger project, shedding light on the textual and extratextual processes involved, and also on the translators' perceptions of these processes, it remains relevant to development studies and other development actors more broadly (Heywood 2019). Although the sample of translators is small, the interviews were extensive and provided rare insight into assumptions around transcription and translation held by translators and end-users alike. Thus we provide recommendations, relevant to those working in both development and translation studies, on how attitudes and approaches could be better articulated to take real-life contexts into account and manage expectations.
This case study exemplifies the informal economy in which many researchers, NGOs and translators operate. It offers a stark contrast to the formalized and heavily theorized translation and research in the Global North where much is digitized, technologically supported and relatively abundantly resourced, and where expectations assume that this extends to the capabilities of others in other sectors and countries. It also foregrounds the male dominance of this case study's informal economy and its multilingual nature. The interviewees, themselves multilingual, living in multilingual societies with oral traditions in which translation is a permanent, daily feature, spoke about their roles and identities in terms of language proficiency rather than translator competency. In contrast to expectations from the north, a professional translator role does not exist, nor was any specific translator training mentioned by any of our respondents.
Consequently, the theorising behind the transcription and translation practices was not convergent and understandings varied considerably. During the interviews, respondents contradicted themselves and each other with regard to translation strategies, working practices and ethics. Many were unclear about the purpose of the task, the nature of the source text or the translation's audience, adopting their "own methodology" and improvising as necessary to mitigate any lack of tools, knowledge and skills. This lack of clarity was compounded by the researchers' erroneous assumption that, having worked in this field for some time under the label of "translators", the participants would know what the task involved and that this would coincide with the researcher's own understanding of it, a faulty chain of assumptions further compounded by the existence of a middleman.
This lack of awareness emerges strongly throughout the article. The translators are not made fully aware of what they have to do or how to do it, and the researchers are not aware of what questions to ask to receive the transcriptions and translations they need. Feedback from end-users to the translators, which can lead to the latter feeling involved and sufficiently empowered to be able to ask questions, is neither offered nor requested. While formulating clear, detailed, even mutually negotiated, instructions may seem time-consuming, they are also likely to prevent the need for work to be repeated when initial translations undermine research aims and are not fit for purpose: in this project, for example, the gender of speakers in the broadcasts was not indicated in some transcribed and translated texts, resulting in initial material being meaningless for a genderbased research. Without feedback loops, improvements to the prevailing ad-hoc and non-systematized situations remain limited.
Translation's essential role in cross-cultural research and development work must, therefore, be recognized; relegating language to secondary importance only contributes to undermining the rigour of the research process. More often, emphasis is on the final output or data set, and the processes of converting communicative texts from one medium to another (oral to written) and from one (or multiple) languages to anotherprocesses on which the whole research and/or development project dependsare sidelined. Yet it must also be acknowledged that the power differentials and examples of social desirability bias, which emerged during the interviews, add to the complexity of the role played by translation. Whilst the researchers (the end-users of the translation), equipped with laptops, recorders, budgets and education, have some power, it is clear that the respondents also wield power; it is they who possess knowledge of the language and can produce whatever form or standard of work they want, as checking or revising it is beyond the researchers' capabilities. Communicating with one another in ways that consider these power relations and facilitate mutual respect and common goals is far from easy. Nonetheless, recommendations do emerge which are aimed at reducing misunderstandings and managing expectations and target all involved.
. Overall, "think about language at the design phase of a project" (Footitt, Crack, and Tesseur 2018, 8). One of The Listening Zone's recommendations for international NGOs, we extend this to a need for recognition by researchers of the primary role of language and translation in development projects. Language and translation emerge as secondary to the main research questions rather than integral to the process. Translation should be included at the start of the project design and not as an afterthought.
. Incorporate into the project design stages for translator selection and the development of networks for future work. The Listening Zone recommends that NGOs "seek to establish a register of translators and interpreters who have worked in and have an understanding of development" (Footitt, Crack, and Tesseur 2018, 8), to which we would add the need to actively consider not only specialisms but local, female capacity, especially in male-dominated informal economies that may not readily present opportunities for women's participation.
. Build in time to consider the people and processes involved in translation. Become familiar with the training, experience and theoretical assumptions of translators and the constraints, conditions and cultural contexts under which the translation is produced, including the available tools and resources. a Ensure appropriate conditions to encourage optimum work. Consider payment per word rather than per job or the provision of an office and equipment. b In the absence of dictionaries or translation memory software, building a shared glossary as part of the translation process will benefit a project that continues over several stages, and can be shared with later projects. The Listening Zone recommends "produc[ing] glossaries of key terms" (Footitt, Crack, and Tesseur 2018, 8), which we extend to include terms and phrases commonly used in the discursive context of the project.
. Work to develop clear and effective feedback loops. a Firstly, researchers must question and be able to clearly articulate what information they want from the foreign language material and what they are using it for. b Secondly, give clear instructions to translators. Discuss the linguistic features, purposes and audiences of the source texts; explain the purpose and audience of the target texts; and discuss which translation methods and strategies will best achieve your research purposes. Orality, for example, may be important, in which case features of orality will need to be indicated. Provide written instructions, explanations and examples, which can be referred to, especially if relying on a middleman. c Thirdly, foster channels of communication enabling translators to ask questions and for constructive feedback to be shared by both parties.

Notes
1. This project was part-funded by the ESRC, UKRI and the University of Sheffield. For project details see https://www.femmepowermentafrique.com/. 2. For the diversity of Tamashek, see Heath (2005). 3. For lists of each country's languages and ethnicities, see Ethnologue: Languages of the World.
https://www.ethnologue.com/ (Accessed April 26, 2019). 4. One example of good practice and the use of glossaries was by Fondation Hirondelle during the 2006 DRC elections. Glossaries were produced by Radio Okapi for journalists to ensure consistent use of electoral terms in the radio's five languages. However, this ceased during subsequent elections (Vuillemin, personal interview, New York, March 2019). 5. All names have been changed.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by Economic and Social Research Council; University of Sheffield.

Notes on contributors
Emma Heywood is a lecturer in the Department of Journalism Studies, the University of Sheffield, UK. Her research focuses on radio, NGOs and women in conflict-affected areas.
Sue-Ann Harding is a senior lecturer in translation studies at the Centre for Translation and Interpreting, Queen's University Belfast, UK. Her main research interests are in social-narrative theory as a mode of inquiry into translations and translated events, with a particular interest in sites of conflict and narrative contestation. ORCID Emma Heywood http://orcid.org/0000-0002-6548-6713 Sue-Ann Harding http://orcid.org/0000-0002-7662-7649