Multilingualism and strategic planning for HIV/AIDS-related health care and communication

Background: Many lower and middle income countries (LMICs) have high levels of linguistic diversity, meaning that health information and care is not available in the languages spoken by the majority of the population. This research investigates the extent to which language needs are taken into account in planning for HIV/AIDS-related health communication in development contexts. Methods: We analysed all HIV/AIDS-related policy documents and reports available via the websites of the Department for International Development UK, The Global Fund, and the Ministries of Health and National AIDS commissions of Burkina Faso, Ghana and Senegal. We used quantitative and qualitative analysis to assess the level of prominence given to language issues, ascertain the level at which mentions occur (donor/funder/national government or commission), and identify the concrete plans for interlingual communication cited in the documents. Results: Of the 314 documents analysed, 35 mention language or translation, but the majority of the mentions are made in passing or in the context of providing background socio-cultural information, the implications of which are not explored. At donor level (DFID), no mentions of language issues were found. Only eight of the documents (2.5%) outline concrete actions for addressing multilingualism in HIV/AIDS-related health communication. These are limited to staff training for sign language, and the production of multilingual resources for large-scale sensitization campaigns. Conclusions: The visibility of language needs in formal planning and reporting in the context of HIV/AIDS-related health care is extremely low. Whilst this low visibility should not be equated to a complete absence of translation or interpreting activity on the ground, it is likely to result in insufficient resources being dedicated to addressing language barriers. Further research is needed to fully understand the ramifications of the low prominence given to questions of language, not least in relation to its impact on gender equality.


Introduction
Effective communication has long been recognised as a crucial factor in health care. Much of the focus of both academic scholarship and training for practitioners has been on intralingual communication, i.e. communication between individuals using the same actual language (English, French, etc.). Scholars have investigated intralingual communication across a wide variety of settings and channels including primary care, telemedicine, media campaigns, online health information, and everyday interpersonal communication (Thompson et al., 2011). There has also been some interest in interlingual communication, i.e. communication between individuals using different actual languages. The bulk of research in this latter area has been on migrant communities accessing health care in Western settings, notably people with limited English proficiency (LEP) accessing health care in the United States (see, for example, Brach et al. (2005); Elderkin- Thompson et al. (2001);Flores, 2005;Jacobs et al. (2004)).
This research has found that language barriers can have a major adverse impact on health and health care (Flores, 2005) and that people with LEP 'have worse access to care, receive poorer quality of care, are less likely to understand and adhere to care plans and are less satisfied with their physicians' (Gregg & Saha, 2007, p. 368). Language barriers have also been associated with additional costs (Hampers & McNulty, 2002), since LEP patients have more diagnostic tests (Brach et al., 2005).
Recent insights into the need for 'transformative' as opposed to 'technical' health communication (Campbell & Scott, 2012, p. 179) reinforces the significance of language: in such transformative approaches, it is necessary not only for the patient to understand the information that is provided by the practitioner, but for the practitioner to understand the patient and to adopt approaches to health communication that are anchored in communities' own analyses of their health needs and problems.
Despite the attested connections between language barriers and effective health care provision, there has been limited research into this issue in parts of the world where language barriers are most acute, in the sense of affecting the largest proportions of a country's population. Many African countries, for example, have extremely high levels of linguistic diversity 1 , with the number of languages spoken within each country ranging from around eight to around 275 (Ethnologue). Although not all scholars agree on these statistics 2 , it is clear that when health care provision is made available only in the official language(s) of the country concerned -usually a single language, and predominantly English or French -then large sections of the population find themselves excluded from good quality health information and unable to be active participants in their own health care, whether at an individual or community level. Whilst individuals -and still less communities -in African countries are rarely, if ever, monolingual, we should note that African multilingualism is 'predominantly oral' (Adejunmobi, 2013, p. 25). This adds to the complexity of the situation, meaning for example that there may be no agreed-upon terminology for many medical conditions or treatments in many of the commonly used languages, or that individuals may not be literate in the languages which they speak or understand most easily.
In the African context, the bulk of scholarly enquiry into language barriers in health care has focussed on South Africa (see e.g. Deumert (2010) (2007)).
The need for further research focussing on African contexts, particularly beyond South Africa, is pressing. Such research will allow us not only to redress the balance of scholarship, which is currently skewed towards Western contexts (Kim et al., 2010), but will also permit a re-envisioning of policy-making to the advantage of those who are currently excluded from health information or health care as a result of language barriers. Since the languages in which health information is exchanged are often those which are acquired via the education system rather than in family settings, addressing policy-making in this regard is also a means of addressing gender inequalities: globally, Sub-Saharan Africa ranks last on the Educational Attainment subindex of the Global Gender Gap, with four of the ten lowest-ranked countries on the literacy rate indicator being from this region (World Economic Forum, 2017, p. 23). This means that women are more likely to be unable to access quality health information when information is provided only in official languages. At the same time, it is women who are disproportionately affected by HIV/AIDS in the Sub-Saharan African region, to the extent that official documents now speak of a 'féminisation de l'épidémie' [feminisation of the epidemic] (CNLS Senegal, 2006, p. 24). Improving communication with women and girls is of paramount importance to efforts to curb infection rates, prevent mother-to-child transmission, and improve the outcomes for those already living with the disease.

Research questions
This study seeks to assess the extent to which linguistic diversity is taken into account in strategic planning for HIV/AIDS health care and communication in development contexts, with a particular focus on West Africa. Where multiple actors are involved, it also aims to identify the level at which strategic planning takes place (donor government, funder, national government, national AIDS commission). Finally, the study summarises the specific means of addressing language barriers that are explicitly outlined in strategic planning documents and reports. While the potential number of organisations spanning these four categories in the 16 countries of West Africa is high, for the purposes of this desk-based research article the scope is restricted to an analysis of one organisation from each of the first two categories, and to three West African government departments with their associated government-led AIDS commissions from the third. The limited internet presence of many of the organisations in the fourth category requires a different set of research methods and merits a separate investigation. The organisations selected for inclusion in the corpus are presented below, together with the rationale for their selection.

Research design
Donor organisation: DFID The UK's DFID represents one of the largest government donors to multilateral funding organisations including The Global Fund (The Global Fund, 2019a), and is also a significant bilateral official development assistance (ODA) provider to African countries 3 . The rationale for its inclusion in this study thus lies in its significance as a global player in the fight against HIV/AIDS: the strategic plans made by DFID not only have direct implications for African countries which receive its bilateral funds, but also potentially influence the strategies adopted by the large funding organisations 4 .

Funder: The Global Fund
The Global Fund is one of the world's three largest AIDS donors (Oomman et al., 2007), and tends to be the major donor in the West and Central Africa region (Avert, 2018). The Global Fund is the principal mechanism used by DFID to finance their contribution to HIV and tuberculosis (TB) treatment (Department for International Development, 2011, p. 12). The rationale for its inclusion in this study thus lies like DFID's in its significance as a donor and major strategic planner in the fight against HIV/AIDS.  Collaborator Network, 2018, pp. 1803-1810. These similarities allow us to hypothesise that our findings will be likely to obtain for other countries in the West African region or even for low-income countries with high linguistic diversity elsewhere in the world. At the same time, the similarities mask important differences. With regard to linguistic diversity, Senegal is the only country of the three (and indeed, the only country in the West African region), to have a national language (Wolof) which has emerged as a national lingua franca and which has come to rival the official language (French) in terms of its use in institutional and educational contexts (McLaughlin, 2008). There are also differences between the three countries in the official language (English in Ghana; French in Senegal and Burkina) and in estimates of the number of proficient speakers of that official language (22% in Burkina and 29% in Senegal (Organisation Internationale de la Francophonie, 2014, p. 17); 61.7% in Ghana ( (Ghana Statistical Service, 2012, p. 6)), as well as in overall literacy rates (36%, 71%, 56% respectively; see Ethnologue). In terms of overall wealth, we should note that Ghana's GDP is more than four times that of Burkina and nearly three times that of Senegal (World Bank, 2018). There are also differences between the three countries in terms of the levels of national government funding allocated to HIV/AIDS spending and the amount of developmental assistance for HIV/AIDS spending received in relation to that state's total HIV/AIDS spending (Burden of Disease Health Financing Collaborator Network, 2018), and slight differences in HIV prevalence rates (0.8% in Burkina, 1.7% in Ghana, 0.4% in Senegal; see UNAIDS). Any of these factors might serve as explanations for differences that may be found in our study, and our research design means that we are not able to isolate one from the other. Nevertheless, the emphasis on qualitative rather than solely quantitative analysis in our study allows us to work inductively to develop further hypotheses which can be tested in subsequent studies.

Ministries of Health and
For each of the three countries, we analysed documents produced both by the Ministry of Health and by the national HIV/AIDS commission which operates under the office of the President in each case. The rationale behind including both types of government organisation is that both are involved in strategic planning pertinent to HIV/AIDS communication, the former primarily in the shape of overarching health policy, and the latter with regard specifically to strategies for targeting HIV/AIDS.

Research methods
Selection of documents for analysis Documents were selected for analysis following principles of systematicity and comprehensiveness. For each of the organisations or government departments in question, we used their official websites to search for all documents relevant to HIV/AIDS. We limited the scope only by document type, restricting our corpus to policy documents (including funding guidelines) and reports, rather than encompassing a broader range of documents such as press releases or research papers. We did not limit the searches by date, including in the corpus any policy documents or reports which the organisation or government department in question made available via their website during the periods in which we gathered our data (November-December 2018 and May-July 2019). In cases where the government website was not functioning, we identified relevant documents through step-by-step processes which are described in detail in the sub-sections below.

DFID
The DFID documents were identified using the UK government website https://www.gov.uk/search/all using the following parameters: Search term: 'HIV'; Topics: all; Sub-topics: all; Show only: 'Guidance and Regulation'; 'Policy papers and consultations'; Organisation: 'DFID'. The search was carried out on 8 July 2019 and yielded 35 results. Of these, six were excluded because their primary focus was on topics other than health (e.g. 'Social Dimensions of Transport') or were not authored by DFID (e.g. pledges made by other organisations to sign up to summit commitments). Some of the results contained child pages, meaning that 66 documents were identified in total through the government website search. One further document, the 'UK Aid Strategy', also available on the gov.uk website, was added to the corpus for the reason that all DFID spending is carried out in line with this strategy.

The Global Fund
The Global Fund website was interrogated between 1 May 2019 and 15 June 2019, and the following documents were selected for inclusion in the corpus: • All documents on the 'Publications' page.
• All documents on the 'Country Coordinating Mechanism' (CCM) page.
• All documents on the 'Funding Guidelines' page.
• All documents pertaining to the three countries under study: Burkina Faso, Ghana, Senegal. These documents were identified via two means: 1) by setting the country filter to the relevant country on the 'Find a Grant' page; 2) via the 'Overview' pages for Burkina Faso, Ghana and Senegal.
The rationale for selecting these documents were as follows: the documents listed on the 'Publications' page can be argued to reflect the key concerns of the Fund and to encompass the most up to date and important strategy documents; the CCMs play a key role in determining the specific strategies used to combat the epidemics; the Funding Guidelines are used by applicants to formulate their requests for funding, and as such shape the projects undertaken; country-specific documentation speaks to the agendas and priorities set by the countries themselves.
The total number of documents selected in this way came to 182. From this initial corpus, any documents which were purely technical (e.g. specimen signature templates), pertained exclusively to topics or regions not relevant to our research or which were shorter summaries of other documents were excluded. After these exclusions, the total number of documents selected for analysis came to 150.  (2018)), and the 'Plan d'urgence pour l'accélération de la réponse nationale au VIH 2017-2018' (mentioned in a UNICEF press release (UNICEF, 2017)). While we did not succeed in finding either of these recent documents, the Google searches did lead to the identification of three further CNLS policy documents.

Ghana: Ministry of Health
We included in the corpus all documents available via the 'policy documents' and 'annual reviews' sections of the Publications tab of the Ministry of Health website with the exception of three documents dealing with topics not relevant to this research (e.g. 'Anti-Malaria Drug Policy'). The date of the interrogation of the website was 3 December 2018.

Ghana: Ghana Aids Commission (GAC)
The Publications page of the Ghana Aids Commission website does not differentiate documents by category. We therefore evaluated each of the 55 documents available on the site, selecting for inclusion in the corpus those which could be classified as policy documents or reports (17 documents in total). The categories of items thus excluded encompassed data management manuals, press releases, and terms of reference documentation. We also excluded abridged versions of documents already included in the corpus. The date of interrogation of the website was 10 December 2018.

Senegal: Ministry of Health
As in the case of the GAC, the Senegal Ministry of Health website does not differentiate documents by category. Once again, we therefore assessed all documents for their relevance to the project, carrying out our website searches between 15-25 May 2019. The vast majority of documents were of a type or on a topic unrelated to our research project (e.g. legal documents, parliamentary bills, instructions on requesting health-related certificates). In total we found eight documents which could be categorised as policy documents or reports. In addition, we used Google to search for the 'Plan national de développement sanitaire ( There was therefore nothing significant missing from the documentation available on the CNLS website and we did not carry out any Google searches for this part of the study.
In the second stage, each hit and its textual context was subject to close reading to ensure the pertinence of the hit for the quantitative part of the study and to generate findings for the qualitative aspect of the study. Close reading was particularly important for guaranteeing the accuracy of the quantitative results, since 'translate' or 'traduire' and their variants are frequently used in both English and French in metaphorical senses rather than to denote translation between languages. For example, in the Global Fund corpus, 'translation' is used in the sense of translation between currencies (2018 Annual Financial Report), translating ideas into practice (Gender Equality Strategy Action Plan 2014-2016), translating issues into investment (Gender Equality and Key Populations), or translating funding requests into disbursement-ready grants (Operational Policy Manual).
'Language' is another term with a broader set of possible meanings than our intended sense of denoting one of the languages spoken in the world. In the corpus as a whole, close reading showed that many of the hits for language denoted intralingual language use, and these were therefore discounted. For example, in the 'Strategic Investments for Adolescents in HIB, Tuberculosis and Malaria Programs Information Note' (The Global Fund, 2016a, p. 15), information is to be provided 'in language that is understandable to adolescents'. This is clearly a reference to age-appropriate language rather than to a particular language such as French or Wolof.

Results
In total, 314 documents were analysed (see extended data (Batchelor et al., 2019) for a full list of documents). The number of documents that contain hits for one or more of the search terms identified above and were deemed through close reading to have mentioned the issue of language use in the sense pertinent to our research question is 35. This equates to 11.1% of the documents. As we will see below, the majority of these mentions are extremely brief.
There are significant differences in the percentage rates for each organisation or government department, as shown in Table 1. Both DFID and the Senegalese Ministry of Health have zero mentions of language or translation. The next lowest in terms of percentage of mentions is the Global Fund. The organisation with the highest percentage of mentions is the Burkina Faso CNLS. Of the three countries studied, Burkina has the highest mention rates (33.3% and 50%), while Senegal has the lowest (0% and 20%). In the case of all three of the countries, the national HIV/AIDS commission is more likely to mention language than the corresponding government Ministry of Health.
While these figures suggest that strategic planning for addressing language barriers is more likely to take place at country level rather than at donor or funder level, in themselves they are too bald Table 1. Number of HIV-related policy documents/reports which mention "language" or "translation" in the sense of addressing language barriers. The data was generated by searching the organisations' official websites between November 2018 and July 2019 and was supplemented by Google searches where websites were not functioning or obviously contained incomplete documentation records.

Number of documents analysed
Number of documents that mention "language" or "translation" Percentage of documents that mention "language" or "translation" to tell us much about the relative prominence given to language issues. They do not differentiate, for example, between background mentions of language on the one hand and sustained discussions of language barriers on the other. In some cases, they also risk being misleading: the four mentions in the Senegal CNLS corpus, for example, all occur in serial annual reports, with exactly the same wording being used on each occasion. Had the corpus only included the most recent annual report, then the percentage calculation would be markedly lower. For these reasons, the sub-sections that follow will present both a nuancing of these quantitative results and a qualitative analysis of the mentions found in the documents.

Global Fund
The majority of the mentions of language in the Global Fund corpus are cursory. For example, the only mention of translation in the 24-page 'Global Fund Gender Equality Strategy' is found in an annex which outlines opportunities for partner engagement: it is noted that 'partners can also provide transport, translation, capacity building and training' (Gender Equality Strategy,p. 20 The Global Fund website lists policy documents and reports under ten different categories. We found significant variations in the number of mentions of language under each of these categories (see Table 2). Table 2 shows that the language issue is accorded greatest attention in the group of documents subsumed under the heading 'Community, Rights, Gender'. The topics with which the documents under this heading deal are community participation, human rights and gender equality, and concern in particular the Global Fund's efforts to ensure that its country dialogue processes are fully inclusive. The tendency for language to be mentioned in connection with country dialogue processes is also found in the documents belonging to the other categories: in all, 20 out of the 25 mentions of language that occur in the 14 Global Fund documents concern community participation in Global Fund agenda-and priority-setting processes. Only five of the mentions of language (occurring across three different documents) address language barriers in the context of health care or health communication activities themselves.
The overall level of attention accorded to questions of interlingual health communication is thus extremely low, and the multilingual nature of many of the environments in which the Global Fund operates emerges only in connection with the Global Fund's operating principles rather than in connection with the actual health care and communication activities that are made possible through the Fund. Indeed, one of the most striking aspects of the Global Fund results is the near-total absence of mentions of language in documentation that is produced  [disparities between sources of information about HIV and the channels through which those living in poverty receive information about HIV. At the national level, the main sources of available information are written material in French (training and information manuals, reports, posters etc.), even though people living in poverty obtain information principally through radio and would rather receive information through this channel.] The way in which this is phrased presents the problem primarily as one of communication channel rather than language, comparing written documentation in French with radio. What is left implicit is that radio communication often takes place in national languages, thanks to a strong network of local radio stations (Capitant, 2008). The fact that the Strategic Framework does not tease apart language and communication channel is an indication of the extent to which questions of language and literacy are interconnected in Burkina: the majority of the population is literate in French rather than in the national languages, and those who have not completed primary school are unlikely to speak French well.  Senegal, 2017, p. 40). The fourth report also introduces the question of the lack of availability of information in sign language (CNLS Senegal, 2017, p. 61) but does not provide a matching initiative for this perceived deficiency.

Discussion
The above results show that the extent to which linguistic diversity is taken into account in strategic planning for HIV/ AIDS-related health communication is minimal. While 35 of the 314 documents mention language, the level of prominence accorded to language barriers is extremely low: most of the mentions are made in passing or are presented as background socio-cultural information, the implications of which are not explored. Only eight of the documents outline any concrete actions for addressing multilingualism in HIV/AIDS-related health communication. This equates to just 2.5% of documents.
The above results suggest that the likelihood of language being mentioned in policy documents or reports is lowest at donor level and highest at the level of the HIV/AIDS commissions of individual countries. If we combine the individual country results and order the statistics by category, this tendency emerges more clearly, as Table 3 demonstrates.
While these differences are striking, it is important to remember that the majority of mentions of language are made in passing, with very few documents outlining or reporting on specific 5 The Handbook asks the readers to consider the following intralingual issues: • Is the word choice clear, or could various audiences interpret it differently?
• Is the language appropriate for your target audience? Obviously you would use a different language when appealing to university researchers than you would when communicating to a youth group. Be careful not to use offensive words.
• Use words that conjure up images and raise emotions, such as empathy for suffering children • Use familiar words the audience will understand and identify with' (Ghana Aids Commission, 2016a, p. 22) Table 4. Number and percentage of HIV-related policy documents/reports which outline concrete plans for addressing language barriers. The data was generated by searching the organisations' official websites between November 2018 and July 2019 and was supplemented by Google searches where websites were not functioning or obviously contained incomplete documentation records. The data is presented by the category of the organisation that produced the document/report.   Table 3. Number and percentage of HIV-related policy documents/reports which mention "language" or "translation" in the sense of addressing language barriers. The data was generated by searching the organisations' official websites between November 2018 and July 2019 and was supplemented by Google searches where websites were not functioning or obviously contained incomplete documentation records. The data is presented by the category of the organisation that produced the document/report.

Category of organisation
Mentions of "language" or "translation" over total number of documents Percentage of documents which mention "language" or "translation" plans for addressing language barriers. When reports that do the latter are grouped by category, it is in fact the international funding organisation that comes out on top, as Table 4 shows: The limited number of countries and organisations included in our study make it difficult, however, to draw anything more than tentative conclusions about the organisational levels at which multilingualism is taken into account. Had we selected a different international funding organisation, for example, it is possible that the figures would look rather different. The same goes for the selection of countries: Senegal, with its very few mentions of language, brings the percentages for the recipient country categories down. It is possible that the statistics would look rather different had we chosen another country in its stead.
It is also important to note that the strategic plans and reports published on the organisations' websites do not tell the whole story. While the Senegal CNLS website contains only one document that outlines specific plans for addressing language barriers, for example, the media library tab on the website contains documentaries and adverts in Senegal's most widely spoken national language, Wolof, as well as the official language, French 6 . From this and other evidence, it is clear that formal health communication activities (in addition to informal, ad-hoc communication between health professionals and patients) do take place in a range of languages in all three 6 See https://www.cnls-senegal.org/mediatheque/films-documentaires/ and https://www.cnls-senegal.org/mediatheque/spots/ . of the countries on which we focussed, but their visibility in formal planning and reporting is extremely low.
Many of the outlines of multilingual initiatives are phrased in general terms and thus make it impossible to ascertain the precise nature of the activities that are envisaged or have been undertaken. For example, the Global Fund technical brief 'Addressing Sex Workers, Men who have Sex with Men, Transgender People…' states that funding requests should ensure that 'the content of behavioral interventions, and of material published in print or online, should be adapted to take into account the needs, culture and language of the key population in question' (The Global Fund, 2017, p. 12). The document leaves open the way in which content and material might be adapted and assumes through its wording that the 'key population' shares a single culture and language.
Where precise activities can be identified in the corpus, they fall into two categories: 1. Training of health care workers. This strategy is mentioned only in connection with sign language. There is no report of any plans to encourage health workers to learn other national languages. This is surprising in the sense that this is a solution that, anecdotally at least, is often adopted by health professionals who find themselves working primarily with patients who speak a language that they do not know; furthermore, it is also a solution that a number of researchers have highlighted as being among the most effective in ensuring effective communication that is built on trust (Claasen et al., 2017).
2. Large scale sensitization campaigns through the production and dissemination of radio or television programmes or adverts in more than one language. While some documents do mention plans for producing written material, in the majority of cases the preferred channel for large-scale communication is audiovisual. This can no doubt be explained by taking into consideration the literacy rates in the three countries under study as well as in many of the countries targeted by DFID or the Global Fund. As we saw above, the conflation of literacy with language proficiency is however not always helpful for thinking through ways to make communication maximally effective.
It is notable that none of the specific initiatives for addressing language barriers envisage making use of internet-or mobile phone-based platforms. This is perhaps a reflection of the lack of sustainability and inclusivity of such initiatives in countries where electricity provision is unreliable and high-speed internet access is limited to city centres 7 , but it is also the case that such initiatives may be going under-reported: in Ghana, for example, there is documented use of teleconsultations by Kumasi Hospital (Molefi, 2010), and other e-and m-health initiatives have been proposed (Osae-Larbi, 2016).
As indicated above, the silence on language barriers that is a feature of the corpus is not matched by a silence on communication in general. For example, the Ghana Ministry of Health's 'Review of Community-based Health Planning Services' makes no mention of language, but mentions communication 13 times (Ghana Ministry of Health, 2009b). Similarly, there is an emphasis throughout the corpus on the importance of community-based approaches and on the role played by community workers and volunteers. The DFID report 'Towards Zero Infections: Two Years On', for example, states that the Community-led TB-HIV Advocacy in Zambia (COTHAZ) project 'trained 100 volunteers […] to provide outreach visits to local communities. Through communicating accurate information, they are challenging myths and stigma surrounding both illnesses' (Department for International Development, 2013, p. 33). It is undoubtedly the case that a key part of community volunteers' importance and their ability to communicate 'accurate information' lies in their language competence: in very simple terms, such volunteers speak the same language as the people in their communities and thus do not face this considerable barrier to effective and accurate communication. It is of course also the case that community volunteers understand and know how to address the culturally-based concerns and preconceptions that might have an effect on the understanding and reception of HIV/AIDS-related information and advice.
The tendency to speak of 'communication' and 'community' rather than to evoke the need for specifically multilingual or language-sensitive approaches might be deemed to be of little importance. After all, it could be argued that policy makers and implementing partners within African countries or other LMICs would be aware of the linguistic make-up of their own countries and therefore not need to spell out the necessity of communicating in a wide range of languages. However, it is almost certainly the case that if language needs are not visible within the funding chain, then they will not have sufficient resources dedicated to them. This, in any case, would be a hypothesis that needs testing, and one which has far-reaching ramifications for the way in which HIV/AIDS-related health funding is allocated.
This hypothesis thus highlights the need for further research in this area. It is not only the findings of this study that need further verification against larger bodies of data and a greater range of organisations and government departments, but it is also crucial that desk-based surveys like the present one be combined with studies of the ways in which health communication initiatives are actually implemented in multilingual societies. Such combined studies will allow us to better understand the impact of the low visibility of language issues on the effectiveness of health communication, particularly in parts of the world that are linguistically diverse or where languages exist in a hegemonic relationship to each other. As noted above, these questions are of keen importance for gender equality: while it has long been recognised that education is crucial in improving the lives and health of women and girls, what has often been overlooked is that one of the key skills that is acquired through education in many linguistically diverse countries is the official language of the country in which girls live. This language allows women and girls to access health care and information within their own countries; if it is a global language, then it gives them access to a large body of healthrelated information; if it is English, the lingua franca of scientific research and the language in which most of the world's public health information is provided, then it potentially allows women and girls to cross the divide between the 'healthinformation haves and have-nots', to use the words of a World Health Organisation report (Adams & Fleck, 2015) 8 .

Data availability
Underlying data All data underlying the results are available as part of the article and no additional source data are required. 1.

Kobus Marais
University of the Free State, Bloemfontein, South Africa The paper studies existing documents from West-African countries on HIV/AIDS that relate to interlingual translation. It collected data from four categories of institutions and tried to ascertain whether these documents mention interlingual translation as an issue in health-care communication. The empirical investigation is set within a theoretical framework that is sensitive to the context in which the research is done, focusing on development issues. The findings are in line with the research problem and flows from the data presented. The ways in which the data was analysed are clear and transparent. I made a number of minor comments in the text: How do you use the term 'West' Africa? Is it just generally geographic, or do are you talking about ECOWAS? Clarifying this might help the reader to understand the context. Introduction, paragraph 6: Also see Chibamba (2018) .
Analysis of documents, paragraph 2: I find this statement problematic. Jakobson made clear that interlingual translation is not the only activity worthy of the name 'translation'. He also included interlinguistic and intersemiotic translation, so I would suggest you reformulate this sentence.
Later in same paragraph: I think we should indeed study all of these as translation. That said, I understand that you need to hedge your choice of data.
References: I am missing references to recent work by a number of scholars such as Footitt, Teseur, Delgado Luchner and even myself. While these are not specifically about health care, the first three do reflect on interlingual translation in the chains of development projects. It might be to the advantage of the authors to take note of these, but it is probably not crucial. That said, I think this is an excellent paper, written with insight and sensitivity, and making a great contribution to translation studies. The findings suggest that minimal attention is given to linguistic diversity across the board, with only 2.5% of the examined documents containing any reference to concrete actions for addressing multilingualism in HIV/AIDS-related health communication. This study thus points to a mismatch between the prevalence of multilingualism on the ground and its lack of visibility at planning and policy level, which has also been documented in other settings, namely in development NGOs and humanitarian organizations.

References
The papers strength resides in its systematic approach, and its attempt to quantify, document and describe a phenomenon that has not yet been studied in depth. The authors disclose the limitations of their study, namely the fact that a different selection of cases might have produced different findings. In this regard, it would have been interesting to provide some information as to why three countries with a comparatively low prevalence of HIV/AIDS were chosen, since a higher prevalence might raise the stakes of HIV/AIDS-related communication.
The study's tentative conclusions are thought-provoking and raise numerous questions that could be explored in subsequent studies. As they move forward in their research, I would encourage the authors to fully engage with the methodological and conceptual implications of the 'predominantly oral' nature of communication in local languages. Indeed, framing multilingualism mainly or predominantly in terms of 'translation' might lead scholars to overlook the myriad hybrid communication practices that exist in African settings, including the wide-spread use of pictographs accompanied by oral explanations in the local language to provide health-care information to illiterate audiences. This aspect could also be explored indirectly on the existing data set by probing for terms such as "sensitization" or "awareness-raising" but also "vocabulary" and "terminology", which in my experience in researching the multilingual communication practices of NGOs are often used as a proxy to refer to settings that involve oral interpretation, sight translation, code-mixing and similar multilingual practices.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Multilingual practices in development NGOs and humanitarian organizations, community interpreting in Africa (Kenya, South Africa).

I confirm that I have read this submission and believe that I have an appropriate level of I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
demonstrate an awareness of the limitations in the scope of their analysis. Given how little attention the topic has received to date, the analysis is nonetheless a good contribution to the topic.
The results of the analysis are, unfortunately, unsurprising, given that similar work in the area of crisis response and the development sector has also found a very limited recognition of the need for translation in policy and guidance documents. Where language and translation are mentioned in the corpus, these mentions are often cursory and/or repetitive, as demonstrated by the detailed analysis of the documents under investigation.
The lack of consideration of multilingual communication is rather shocking when the context of multilingualism and (sometimes) low literacy is considered. This is symptomatic of a much larger blind spot on the importance of language and translation across governmental organisations and NGOs internationally. As highlighted by the authors, the lack of mention of language is problematic in itself, but lack of mention in policy and guidance documents also implies a lack of action, which is of a much greater concern, given the potential ramifications.
The authors call for more research on the topic, but also rightly point out that the need really lies with investigating the actual implementation (and the success or failure) of communication campaigns on the ground in highly diverse multilingual settings. While a call for more research, and especially field research, is welcome, the authors might have gone a few steps further by making concrete policy recommendations for inclusion in the types of documents they analysed.
There were a few minor aspects of the article that could be improved. For example: In the Introduction, para. 4, a statement is made that when healthcare information is provided in a single language this is "predominantly English or French". This statement needs to be supported by referring exactly to which parts of the globe this pertains.
On page 4, where the justification is given for the inclusion of specific Ministries of Health, it would be useful to have a footnote providing an explanation of what the linguistic diversity indices typically mean.
It was a little surprising that "interpreting" and derivations of that term were not listed as a search term, given the role of interpreting in health communication and the point made about the dominance of orality in the African context. It would be useful if the authors addressed this issue. Presumably, on close reading of the documents, it was found that there were no occurrences of such terms?

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? Not applicable Are all the source data underlying the results available to ensure full reproducibility?

Yes © 2019 Footitt H. This is an open access peer review report distributed under the terms of the Creative Commons
Their conclusions are striking. They conclude that the extent to which linguistic diversity is taken into account in strategic planning is minimal, with a suggestion that the likelihood of language being mentioned is lowest at donor level, and highest at the level of in-country commissions. The silence on language barriers which they detect is not matched by a similar silence in these documents on communication in general, or on the desirability of community-based approaches. Whilst it might be argued that partners and policy-makers within the African countries can be left to address the diversity of language needs, this apparent lack of any overt recognition of multilingualism throughout the policy-making chain suggests that the resources needed to translate and interpret in these highly diverse linguistic contexts are very unlikely to be provided.
Overall, my only suggestion which the authors might consider is to strengthen the initial part of the article with a slightly fuller description of current work on the problematics of migrant communities assessing health care in Western settings where conclusions on access, quality of care, and costs associated with languages provide helpful pointers for the article's broader considerations of multilingualism and health care in the African context. This is an original and carefully written article which strongly points to the need for further research on the ways in which health initiatives are actually implemented in multilingual societies, how the actions of health policies are actually translated into the words of the receiving communities.
Is the work clearly and accurately presented and does it cite the current literature?