Knowledge Translation in the Global South: An Exploratory Mapping of the Literature

This paper maps the literature that focuses on knowledge translation (KT) in the global South. It was commissioned as part of the KT in the Global South research project, supported by Canada’s International Development Research Centre (IDRC) and led by the Institute of Development Studies (IDS). The trilingual systematised review helps discern where information about KT is missing, emerging, or well-established, and highlights information on what the KT strategies employed are, where, how, by whom, and for whom.

The knowledge base generally provides good coverage on many aspects of KT, even as the scope of individual studies is extremely variable, and some areas remain under-researched (e.g. Asia and the Pacific, rural areas, contexts of war or autocracy, longer study time frames). In particular, the knowledge base amounts to a large coverage regarding: the aggregate size of studied populations, groups, or samples; geographies; types of political, economic, social, and cultural contexts; and time frames. There is also good, though differentiated, coverage about the main areas of change in processes and outcomes that KT may aim for (namely policy, practices, and collective action), about the causes, dynamics, and effects of both KT and the change that KT may aim for, as well as about the actors supporting KT, the actors targeted for change, and the types of knowledge being 'translated'.
Differences between the global North and South are primarily described as differences by stakeholders, with some minor coverage of differences by intended outcomes. Looking at practices as documented in publications, many researchers and research institutions practice KT only to a limited extent, using a narrow range of easier, less interactive KT activities with limited attention to target audiences. Often knowledge management has remained oriented towards

Acknowledgements
This report was produced as part of the KT in the Global South research project, funded by Canada's International Development Research Centre (IDRC). IDS wishes to thank IDRC for its financial and intellectual support. We are also grateful to those who provided peer review of this publication and particularly want to thank the members of the project's Steering Committee who informed our approach.

Introduction
The movement of engaging research with policy and practice goes back several decades and although its roots sit firmly within the health sciences, it has become an important feature across disciplines, sectors, and geographies . A commonly used term to describe these processes is 'knowledge translation' (KT) (see Box 1.1). However, while there is a wealth of literature that has established 'knowledge translation', particularly in health research in highincome countries (Boaz, Fitzpatrick and Shaw 2009;Farley-Ripple, Oliver and Boaz 2020;Oliver et al. 2022), the discourse about KT in the global South and low-and middle-income countries seems to be less well established.
A lot of research and work has been conducted to investigate the use of knowledge or evidence to inform practice, policy formulation and implementation, with authors discussing the nature of these processes (ontology) and theory of knowledge (epistemology). For instance, Graham et al. (2006) provide a categorisation and a conceptual framework for thinking about the process and integrate the roles of knowledge creation and knowledge application. 1 Furthermore, different forms of knowledge translation, exchange, transfer, and brokerage between research and policy and practice are described (Graham et al. 2006).

Box 1.1 Choice of terms: 'knowledge translation'
Numerous terms are used by KT actors and by those who study KT to describe all or part of the process. The terms have varied over time, differ between sectoral, professional and thematic fields, and are expressed diversely in different languages. For lists of frequent terms and discussions of these, see for example: and cultural factors that promote or limit knowledge use (Punton 2016). This encompasses interventions that are institutionalised, with systems-level support designed to enable knowledge use within specific policy contexts. This might include training for policy actors in specific ministries, research policy partnerships focused on challenges, or the creation of new units or bodies designed to synthesise knowledge and evaluate programmes (Georgalakis 2022). This mode can include elements of all the other modes, as well as research capacity-building and knowledge translation, science advisory services and knowledge brokering.
Irrespective of these overlapping and interconnected modes of knowledge translation, the impact of knowledge on policy and practice remains variable and difficult to unpack, often taking time and indirect routes.
Such knowledge gaps on KT have particularly serious repercussions in the global South. In global South contexts, the need for evidence-informed solutions to development challenges is acute, but inequities in the production and use of knowledge, paired with socio-political constraints, often hamper the possibilities of generating new research or utilising existing knowledge (Global Commission on Evidence to Address Societal Challenges 2022).
This is illustrated by studies on influencing policy with research in the global South (there are analogous findings on influencing practice with research). For instance, a cross-regional and cross-sectoral study, developed at the UK-based Overseas Development Institute (ODI), compared 50 cases of successful or failed influence of research on policy about development issues in the global South and countries in transition Young 2003, 2006). Results showed that the influence of research on policy was sometimes immediate and direct, but in most cases it 'was less direct and took some time, requiring strenuous advocacy' (Court and Young 2003: vii). Moreover, both foreign and domestic factors had a major influence on the domestic links between research and policy in the global South. For example, multilateral funding organisations typically influenced policy contexts and research priorities, raising concerns about priorities, legitimacy, and ownership (Court and Young 2006).
Similarly, another cross-regional and cross-sectoral study compared 23 cases of researchers trying to influence policy in diverse sectors and countries in Africa, Asia, and Latin America. It showed that researchers can influence public policy decisions while policymakers influence policy decisions in the global South with timely and relevant knowledge, but that successful influence depends on a combination of policy context and deliberate KT strategies (Carden 2009).
Both Northern-produced syntheses (Carden 2009;Young 2003, 2006) argue that the differences in contexts between the global North and the global South are so great in degree, and at times in nature, that findings about KT in the global South are markedly different. Differences arise due to varying political contexts that in turn limit the applicability of policy processes. Theories of the global North policy processes and the percolation of ideas do apply, but they are insufficient and fail to address the political complexity of the global South Young 2003: vii, 2006: 456).
Policy change is described as a function of political demand and political contestation. Within this, critical aspects that shape the links between research and policy are, in particular: demand for change and new ideas by policymakers and society, the nature and degree of political contestation, and openness to new ideas, which also relates to the prevailing discourses among policymakers Young 2003: vii, 2006: 456). In adverse contexts, frequent political resistance to change was a serious obstacle, even where there was clear evidence. In supportive contexts, political openness and a favourable political culture enabled the use of research in policymaking (Court and Young 2003: vii).

Box 1.2 Choice of terms: 'global North' and 'global South'
Any terms to describe the global North and the global South are contested and come with conceptual, theoretical, ethical, and political limitations. There are large bodies of literature offering solid critiques of related terms such as the following (which are not synonymous): 'developed vs developing countries', 'high-income vs low-and middle-income countries', 'the West and the rest', 'First World vs Third World', 'former colonial powers vs post-colonial countries', 'core vs periphery', and 'global North vs global South'. One shared problem is that these categorisations are binary and so broad that they are inevitably oversimplifications, grouping into one binary element countries that have major differences in the profile of their development and in their histories, societies, and economies. Alternatives that spell out lists of specific countries and regions quickly become unwieldy if they need to be used repeatedly in writing.
Consequently, by agreement among the project participants, this research refers to 'global North' and 'global South' as the least unsatisfactory terms. The authors acknowledge that this terminology still has limitations and is problematic in some respects, among others because its broad grouping of countries erases important differences within each category, and because it is vague about which countries get assigned to which side of the binary. But the comparative advantages of this terminology make it a good choice as a working tool. See its use by United Nations (UN) agencies such as the UN Conference on Trade and Development (UNCTAD), for example in its 2018 report Forging a Path Beyond Borders: The Global South, and the UN Office for South-South Cooperation (UNOSSC), supported by the UN Development Programme (UNDP).

Rationale of the present review
Given the historic overreliance on Northern-based English-language literature in this field, the persistent lack of attention to power, and to the plurality and complexity of political contexts, sectors, knowledges, and actors involved in KT (Jones et al. 2009) -as well as the possible differences in the applicability of findings originating from the global North to the global South -the present trilingual literature search and mapping aims to explore the literature in and from the global South. The aim of this paper is to explore strategies used to enhance the use of knowledge in Southern contexts of policy or practice.
This endeavour acknowledges that the KT literature seems to have made only a modest difference in actually improving how knowledge and research influences policy or practice for development (see, for example, Olivier de Sardan 2021). Scholarship in this field often fails to connect with the experiences of those working at the interface of research organisations and policy communities (Georgalakis 2021). This is particularly true for those working outside of health and beyond high-income countries. The concepts used to describe the use of knowledge in policy or practice can simply seem too abstract to be of any practical use, and endless discussions around how to define terms such as knowledge translation, or research mobilisation and uptake, often fail to grasp what is really going on. 2 This working paper firstly screened both academic and grey literature to determine an inclusive search strategy that can be applied to discern information about KT in the global South. The paper aims to provide information on: 1. The primary strategies used to enhance the use of knowledge in Southern contexts.
2. How these strategies differ in relation to key stakeholders and intended outcomes (changes).
3. How the different strategies relate to sectors, disciplines, and geographies.
4. How the different modes of KT interventions (Figure 1.1) reflect different theories of evidence-informed policy and practice.
The review is thus meant to help discern where information about KT is missing, emerging, or well-established, and to highlight information on what the KT strategies employed are, where, how, by whom, and for whom.

Methodology
The present review is based on a mix of review methodologies and can be characterised as a systematised review (Grant and Booth 2009)   Searching with the exact expression 'knowledge translation' in the three languages resulted in numerous publications but with a narrow scope. Most were in English, from and about the global North, and concerned the health sector. This also confirmed that while practices of KT do exist in diverse contexts and forms worldwide, the vast literature about KT does not necessarily refer to it as 'knowledge translation'. Therefore, search terms were broadened and tested. An iterative process including KT stakeholders from the global South was employed. First a combination of keywords about 'knowledge' and 'translation' yielded in an enormous quantity of results, a number of them irrelevant. Some of the keywords were too general ('knowledge') by themselves, and keywords with multiple meanings lead to partly irrelevant results (for example, 'translation' also led to results about linguistic translation). Furthermore, initial search results showed that the expression 'knowledge translation' was not used in a consistent way by different actors. For example, some organisations included technology transfer or commercialisation under its umbrella, whereas others did not (see, for example, Straus et al. 2009). This meant that using the exact expression would limit finding relevant literature, but that broadening search terms too much would swiftly become unmanageable. Hence, it was important to use targeted combinations of keywords that would strike the right balance and elicit enough, but not too many, relevant results.
The initial searches also revealed that 'knowledge translation' does not automatically translate well, both between languages and between sectors or thematic issues. There are variations by sector, discipline, issue, language, and geography. Consequently, the trilingual searches required translating keywords from English into the other two languages, but also required cultural adaptation to their context. For example, in French, a number of results on 'traduction des connaissances' are about translation between languages, and it can be important to also use keywords relating to 'communication', 'vulgarisation' and 'médiation'. In Spanish, variations of the expression 'traducción de conocimientos' get fewer results than, for example, 'aplicación' or 'comunicación', but more than 'divulgación ', 'socialización', or 'transposición'. Semantic differences between sectors were addressed in a similar way. While some sectors, such as health, appear to frequently use the exact expression 'knowledge translation', others do not and either use related synonyms or entirely differently concepts (for example, in the education sector, 'teaching', 'learning', and 'pedagogy' often cover aspects of KT, while in agriculture, 'extension services' include some KT).
In order to diversify the search results, keywords about 'citizen' were included as well. This provided a wider spectrum of references in terms of sector, geography, and types of actors involved.
The final search terms included the exact expression 'knowledge translation' and immediately related expressions, with broader synonyms and associated terms on both 'knowledge' and 'translation' added in a managed way. Therefore, literature on KT was searched by using the exact expression 'knowledge translation', and combinations of synonyms about knowledge (such as 'research', 'evidence', or 'lessons') and KT (such as 'evidence-based', or 'knowledge + policy', 'action', 'practice', or 'citizen').
A strategic decision to proceed in this way was made because academic database Web of Science can run complex search queries with a large number of search terms. Depending on the number of results returned, search strings were then narrowed down as needed by taking out some broad keywords that brought in too many results, or by adding in sets of synonyms about one or more of the following three thematic areas: development or aid; global South; and/or reviews (as in 'literature review', to retain a manageable number of references if initial searches resulted in too many references).
The final search terms used are summarised in Table 2.2, which presents the most frequently tested and used search terms -there were adjustments and variations depending on the source searches (Web of Science, Google Scholar, or Google) and on the results returned, but Table 2.2 offers an overview.

Approach to reference selection and analysis
The final literature searches were run in March 2022 in Web of Science (Web of Knowledge) and in March and April 2022 in Google and Google Scholar. Web of Science was chosen because it covers a wide range of disciplines -from medicine and natural sciences to social sciences and humanities -and various journals within each discipline.
In addition, a non-systematic snowballing from a few of the publications found through the structured searches was used, examining the reference list of the given publication or the web pages associated with it to examine additional publications that did not come up in the systematic searches.

Criteria for inclusion, organisation, and prioritisation of references
References were considered eligible for inclusion if they had an empirical focus on KT in the global South. The concept of 'global South' was approached in an open-ended way and included anything labelled as 'global South', as well as all countries other than high-income ones, and countries and territories with middleor low-rank socioeconomic indicators in sub-Saharan Africa, Central and South America and the Caribbean, Asia and the Pacific, the Middle East (Western Asia) and North Africa, and the Caucasus and Balkans. There was no restriction on sectors included in the review. References that covered theoretical debates or conceptualisations of KT and/or were clearly related to the 'global North' were not included in this review.
Owing to the multitude of references found in the preliminary searches, the present paper only investigated publications from the past five years . Academic publications included scientific journal articles, proceedings, and literature reviews (all peer-reviewed), while grey literature included publications such as reports, case studies, policy briefs, guidance notes, manuals and capacity-building materials, guides on resources, meeting summaries, periodicals such as bulletins, books, working papers, presentations at conferences or workshops, and web pages, including blog posts. The review did not include programme-related publications, such as programme reporting, annual reviews, and programme evaluations.
When a publication was available in more than one language, the first choice was English followed by French and Spanish, unless there was a notable difference in the presentation that led to a clearer, more precise description in one of the other languages, in which case that version was selected (for example, if an English translation was not of very high quality). All references retrieved were subsequently categorised through a structured approach. Four main descriptors were used for the categorisation: (1) geographic coverage (single vs more than one world region), (2) sectoral coverage (single vs more than one sector), (3) focus on inequalities (significant focus of the publication on specific groups or structural inequalities -such as gender -vs no such focus), and (4) focus on specific KT actors (significant focus of the publication on specific KT actors -such as thinktanks or non-governmental organisations (NGOs) -vs no such focus).
Within each of the above categories, the descriptors of the other three categories were also applied as much as possible -with priority to geography and sectors as major categories. For example, a reference whose content revolved around focusing on a specific sector was first dispatched to the section on sectoral references, and within it to the subsection on the applicable sector. Then, within that subsection, the reference was assigned to the subsection about the geographic area it covered. Eventually, the bibliography got structured into four sections based on those four descriptors, as well as a section for cross-cutting references that cover multiple world regions and multiple sectors, i.e. references with the greatest breadth and diversity of scope.
The present paper did not apply a quality assessment of the references but instead used prioritisation criteria to determine whether an included reference would be ranked as high, medium, or low priority in the organisation of the bibliography. Priority criteria were set in two steps. The first priority was a focus entirely on the global South (rather than a coverage of the global South alongside the global North 4 ). This was followed by having a cross-cutting coverage: the highest priority was given to references that had a multiregional and multisectoral focus. This was followed by references that comprised a narrower geographical focus and fewer sectors, and single-country case studies received the lowest priority.
To ensure some basic quality control, publications were assessed and prioritised based on very basic criteria of methodological rigour, namely: offering a description of the methods used to make the empirical claims, and having some form of basic peer review or editing (for example, academic peer review or publishing, or editing by an organisation or publisher on non-peer-reviewed literature led to a higher priority than an academic conference paper or a selfpublished opinion piece). We did not apply quality scoring for each publication. 4 This was in line with the focus of this review. Additionally, one practical advantage of proceeding in this way was that references with joint North-South coverage often presented findings about the South merged with those about the North, or in ways that made it hard or impossible to distinguish.

Data extraction and analysis
Publications were mapped as described above to create a structured bibliography and reviewed in a narrative, exploratory way. Grouping and categorising of the references consisted of four strands of thematic analysis. The first was the thematic grouping of references used to construct the structured bibliography. The second was an in-depth description and mapping of the state of knowledge, as observed during literature searches and as shown in the final structured bibliography. The third investigated the themes that emerged during the narrative content exploration, in connection with the four research questions. The fourth was the themes that emerged not directly in relation to the research question, but from references that brought to light interesting aspects of KT (e.g. on equality and intersectionality). The approach, therefore, combined deductive components, relating to the research framework, and inductive components, based on themes that emerged during the literature searches.
The paper will continue with presenting and discussing the state of knowledge in the global South. Results will be discussed as part of the knowledge presentation. The paper will then present reflections on the literature, conclusions and limitations of the research approach.

Search results
In total, 202 references were included in the mapping of the literature. The structured bibliography can be found in Annexe 1. Most references -87 per cent (n = 176) -were written in English, with 4 per cent (n = 8) in French, and 9 per cent (n = 18) in Spanish. In total, 152 references (75 per cent) were scientific references, such as research publications (n = 87), literature reviews (n = 62), scientific journal editorials (n = 2), and Master's Thesis (n = 1). Furthermore, 14 references were books and one a book chapter. The grey literature amounted to 35 references encompassing reports (n = 24), working papers (n = 5), discussion papers (n = 2), and others such as rapid response documentation, manuals, and seminar write-ups (n = 4).
References were categorised into seven main categories. Most articles (n = 83, 41 per cent) were categorised as cross-cutting references, which do not have a specific geographic focus. Of these, 23 (28 per cent) also did not have a specific sectoral focus, while 32 (39 per cent) of the references were focusing on health topics. Furthermore, ten references (12 per cent) concerned environmental topics in several regions and seven (8 per cent) education-related topics. A further five references (6 per cent) addressed information and communication technologies and four references (5 per cent) were related to agriculture. The remaining references presented topics related to infrastructure and hygiene.
Four categories were created representing geographical areas. In total, 30 references were related to topics in sub-Saharan Africa, of which 19 (63.3 per cent) concerned health and 36.7 per cent (n = 11) were references that did not have a specific sectoral focus. The second category comprised references that address topics in Latin America and the Caribbean. This category consists of 15 references, of which six (40 per cent) do not have a specific sectoral focus, six (40 per cent) are health related, and three (20 per cent) relate to agriculture. There were two references that addressed health-related topics in the Middle East and North Africa region, and eight references that address topics in Asia and the Pacific region. Of these, seven references were related to health.
Another category of references centred around intersecting factors. In total, 25 references were categorised into this group. References addressed topics such as disability (n = 8, 32 per cent), age (n = 6, 24 per cent), race (n = 2, 8 per cent), and gender (n = 2, 8 per cent). The remaining references focused on health, education, or did not specifically address one topic. The last category comprises references that concern different type of KT actors. Most (n = 12, 31 per cent) of these references relate to universities and research institutions. Furthermore, nine references (23 per cent) relate to knowledge brokers and seven (18 per cent) to KT networks. The remaining references address KT by grass-roots organisations (n = 5), research funders (n = 4), NGOs (n = 2), KT by end user (n = 1), and art (n = 1).

Methodologies used in the knowledge base
In general, references spell out their rationale, methodology, limitations, and findings. Furthermore, most references acknowledge at least some of the epistemological, ethical, and political issues associated with KT in general ( There is coverage of major areas of change in processes and outcomes that KT may aim for, namely policy, practices, and collective action such as social mobilisation, organised citizen involvement, education, or participation. There can of course be some overlap between these areas.

Size of studied populations, groups or samples
Taking all the included literature into account, the aggregate size of studied populations, groups or samples is large, but their size in individual studies is extremely variable, from large studies (e.g. CDKN 2021; Erismann et al. 2021;Georgalakis et al. 2017;Goldman and Pabari 2020;Manning, Goldman and Hernández Licona 2020;Shroff et al. 2017, cited in Murunga et al. 2020Oliver et al. 2021;UNESCO 2021) Taremwa et al. 2020;Uneke et al. 2017). While smaller sizes of study are not a problem in themselves, they do limit the generalisability and, in some cases, the power of the findings (for example, in the case of quantitative studies that apply inferential statistics to small purposive samples).

Time frames
References encompass various time frames, though publications with more limited time frames seem somewhat more frequent. A greater number of studies refer to one-time cross-sectional assessments or data, or short-term effects In particular, some studies have very narrow time frames, such as a one-off survey and consultation of 40 participants in a one-day event for stakeholders' engagement in Nigeria . With such studies, the generalisability and sustainability of findings are typically unknown. By contrast, other studies follow developments in KT over months, years, or decades (see, for example, a study covering two decades, Majdzadeh et al. 2010, cited in Murunga et al. 2020, and a book synthesising research spanning four decades, Olivier de Sardan 2021). The longer the time perspective, the more comprehensive the picture of KT factors, practices, dynamics, and effects, be they negative, neutral, or positive.

Diversity and balance in the knowledge base
Search results show that there are various academic, practitioner, and policy sources producing relevant publications. This diversity strengthens the overall rigour of the knowledge base, since these three types of sources typically approach the study of KT in the global South with different purposes, questions, and methods from one another.
Compared to practice-and policy-oriented literature, academic literature tends to view KT more holistically. This is connected to academic sources often embedding their perspectives in considerations of history, politics, society, or economy (e.g. Adu  A systematic review about KT found similar results, noting that most of the 66 original research papers it included were case studies (Murunga et al. 2020).
Furthermore, the type of material varies, with a mix of primary and field-based research (e.g. in case studies), and secondary and desk-based research (e.g. in literature reviews knowledge across domains. It views knowledge as a global public good and a part of the commons. 5 It embraces the idea of multiple knowledges, of emergence and complexity in knowledge processes, and emphasises the importance of local knowledge. This approach typically involves multistakeholder processes (Cummings et al. 2019). The authors suggest a framework for systematic knowledge management (Table 3.2) consisting of three components. The first is based on practice and encompasses two concepts of embedded knowledge, both of which typically face funding constraints. The first concept relates to intra-organisational processes and its usual tools are information and communications technologies (ICTs), audits and scans for knowledge management, reference to best practices and case studies, and peer assistance. The other concept relates to inter-organisational processes, through communities of practices between organisations. Its usual tools are ICTs (Hernández-Soto et al. 2021: 239) -with a particular role for social media -as well as reference to best practices and case studies, and peer assistance.
Knowledge exchange within such virtual communities of practice (VCOPs) is examined in an exploratory systematic review of English-language literature, though the review was published in Spanish. So far, the mechanisms of participation in VCOPs, especially knowledge-sharing behaviour, have been studied by different authors based on extremely heterogenous theories and methods, leading to the identification of a wide array of factors and dimensions.
To synthesise such knowledge, the review brought together findings from academic literature about the key drivers of knowledge-sharing behaviour in VCOPs. The review focused on literature published since 2010 and was based on 42 studies and a qualitative content analysis. Usefully, the review included references that were diverse in their methodologies (quantitative, qualitative, and mixed), the VCOPs' participants (businesses, education, health and technology), and geographies ( The second component in the framework is societal. It centres integrating and co-creating knowledge across domains, and emphases the ecology and ecosystems of knowledge. As previously mentioned, it refers to multiple 5 'The commons' is the cultural and natural resources accessible to all members of a society, including natural materials such as air, water, and a habitable Earth. These resources are held in common even when owned privately or publicly.

Structures of inequalities and intersectionality
There is some coverage on structures of inequalities and their intersectionality in KT, even though much of the literature on KT in the global South explicitly aims to tackle development issues related to poverty and other structural inequalities. 6 6 Structures of inequalities include: socioeconomic class (including wealth, income, education, position in relation to capital and to paid and unpaid labour); caste; gender and sexualities; race, ethnicity, culture, Publications sometimes refer to inequalities as part of the knowledge being translated. For example, the knowledge being translated may use genderdisaggregated data or explicitly refer to gender issues (e.g. Cameron et al. 2020;. But they rarely examine inequalities as part of KT itself.
Very few references about KT focus on one or more structures of inequalities in KT altogether (please see Annexe 1: Structured Bibliography). For example, very few publications are dedicated to gender issues in KT in the global South (exceptions include Ng 2020; Rasekoala 2019; . As for the bulk of literature on KT, it has no focus on inequalities, and largely fails to mainstream the consideration of structures of inequalities, let alone intersectionality. 7 Within the literature that does address inequalities, there is very uneven coverage of different inequalities, with serious gaps about some inequalities and about important facets of specific inequalities. For example, whilst there is some work related to disability (e.g. , there appears to be a dearth of work on caste. But even within available publications on disability, there are serious problems, including the adoption of a largely medical approach to disability instead of due consideration of the social model of disability alongside a medical model where applicable; and a focus on physical disabilities, with little on sensory disabilities (such as deafness or blindness), intellectual disabilities (such as learning and communication disabilities, neurodivergence), and mental or psychological disabilities (such as schizophrenia, complex trauma, depression, or anxiety).
Furthermore, within individual references, coverage of all major types of inequalities is not systematic. For example, a publication may cover age and gender, but no other structure. Even when references consider structural inequalities, a number of them often use weaker approaches, in that they use flat descriptions of situations instead of analyses of structural inequalities as relations of power; for example, referring to 'women and girls' as a stand-in for 'gender'. Typically, references will use descriptions of status in hierarchies and of specific groups, instead of more comprehensive analyses of power relations that would also include analysing the power dynamics that connect the different groups, from exclusion to adverse inclusion and exploitation. and religion; nationality, indigeneity, migration or displacement status; geography (including urban or rural setting, remoteness, and physical and service infrastructures); age; health; and disability. Most references fail to adopt an intersectional approach. Among the references that do discuss more than one structure of inequality, inequalities are approached simply as additional to one another, for example through crosscategories such as 'under 18 and indigenous'. There often lacks attention to the compounding dynamics of inequalities, and to their particular contextual configurations in a given time and place.
Even when references consider power relations, many still tend to adopt approaches that are 'aware of', 'sensitive to', or 'responsive to' inequalities, rather than transformative, to borrow language from gender studies (among the exceptions are, for example, Ng 2020; Rasekoala 2019). 8

General factors influencing KT practice
Murunga et al. undertook a systematic review of KT by researchers and research institutions in LMICs, which included 40 articles, and synthesised factors that influence the practice of KT by researchers in LMICs (2020: 9, 11). The review highlighted that research use by target audiences was not commonly reported and was only mentioned in a few studies. Context-specific factors were also mentioned, such as researchers' interest in KT and institutional incentives promoting KT. Furthermore, researchers' and their institutions' credibility and reputation, and the relevance of research evidence, as perceived by target audiences, 9 were other factors discussed. The review by Murunga et al. (2020) also mentioned factors related to researchers' contextual understanding, and decision makers' research background, as well as the relations, incentives, and stances between researchers and policymakers. For example, the differences in 'views, demands and incentives among researchers and policymakers in relation to research, its use, policy actor roles and policymaking' (ibid.: 11). Issues such as the nature of the policy issues (for example, whether the issues are seen as technical, contested, or in the policymakers' interest) and power in situations, structures, and systems which highlight the political context, donor influence, and international influence, such as the influence by the World Health Organization (WHO) on decision-making.

Differences by stakeholders and intended changes
The review of the KT literature shows a greater focus of references around the types of stakeholders, with limited analyses organised around the types of 8 As examples drawn from the vast literature about the spectrum of approaches to gender, see two practitioner-oriented guides (Hillenbrand et al. 2015;UNICEF n.d.) and one academic publication (Pederson, Greaves and Poole 2015). 9 In exchanges with the paper authors, Enrique Mendizabal noted that this factor has already been well studied, as evidenced by work about the credibility of thinktanks conducted by On Think Tanks (OTT). This includes a working paper and a reading list of English-language publications by OTT. intended outcomes, such as policy vs practice vs mobilisation, or detailed outcomes (for example, in policy, the difference between aiming to influence ideas, values, interests, processes, institutions, structures, or decisions). 10 Coverage is good but differentiated on the causes, dynamics, and effects of both KT and the change that KT may aim for. It is likewise good but differentiated on the stakeholders examined, in their nature -individuals or institutions -and types (e.g. state officials vs private pressure groups vs NGOs vs general public). For example, some references cover a wide array of targeted actors (e.g. Oliver Among publications that focused on one or more specific types of KT actors, the present research found references about a wealth of actors: universities or academics (individuals or journals); research funders (including funding institutions); aid donors; knowledge brokers; NGOs (including through engagement with evaluations); grass-roots or community-based participation (informal, or more formalised such as community advisory boards); media; end users' online searches; multiple stakeholders interacting through institutions, partnerships, networks, or secondments; artists; the state and public administrations; and private businesses (see Annexe 1 for the respective references).
The references included in this review show that there is a disparate coverage for types of activities and interventions for KT. A significant number of references focus on the efforts at KT made by knowledge producers and brokers (sometimes called the 'supply side') (e.g. Díaz Larrañaga et al. 2019;Etomaru et al. 2021;Kalbarczyk et al. 2021b;Mc Sween-Cadieux et al. 2017;). There seem to be far fewer studies on the knowledge users' side ('demand side') (e.g. Stewart, Erasmus and UJ-BCURE Team 2017;Uneke et al. 2017). There seems to be a growing number of studies on relationships between knowledge producers or brokers and knowledge users (e.g. Etale and Jessani 2020;Georgalakis et al. 2017;Jessani et al. 2021;Keita et al. 2017;Mpando et al. 2021;Nguyen et al. 2020;Oliver et al. 2022;Oliver et al. 2021;Roe and Stanojević 2020;Uneke et al. 2018). 11 As for the types of knowledge dynamics that are studied, there seems to be distinctly more research respectively on formal KT initiatives and 10 This is in contrast, for example, with vast literature in political science that is not specialised in KT, which offers studies organised around stakeholders but also around policymaking processes or outcomes.

11
The categories of knowledge producer, broker, and user can of course be fluid and overlap (for example, in the more interactive or grass-roots types of KT, someone can take up all three roles, alternatively or simultaneously). Nonetheless, the categories are still useful for identifying people's and organisations' positions in relation to knowledge and to KT. institutions than informal practices, on top-down KT than bottom-up KT, and on directed or facilitated KT than diffuse or self-led KT.
Overall, the diversity of research on the types of knowledge being translated and on the actors supporting KT ensures that all main categories of knowledge type and KT actors are covered, although the degree of research is uneven. In particular, this review found distinctly more research that revolved around academic knowledge and academia-related actors of KT (e.g. Murunga et al. 2020) than on any other knowledge type (such as informal knowledge, or scientific research from sources outside academia) and any other KT actor (such as thinktanks, NGOs, or grass-roots movements, for example grass-roots community advisory boards who translate knowledge in 'vernacular' ). This likely relates to two issues: (1) academic research using the exact term 'knowledge translation', and (2) the exact term 'knowledge translation' being used less in non-scientific settings and outside the health sector.

Researchers and research institutions
The aforementioned coverage patterns result not only from knowledge producers' and brokers' decisions, but also from the actual state of KT practices around the world -especially more formalised KT. For example, Murunga et al. (2020) conducted a systematic review documenting and analysing KT practices and activities undertaken by researchers and research organisations. This review found a small knowledge base with limited sectoral diversity, mainly focusing on health, and documents and analyses KT practices in predominantly LMICs. Furthermore, about half of the researchers in LMICs do not practice KT (Murunga et al. 2020). Instead, researchers and research institutions allocate more effort and investments to roles and functions in research than in KT (ibid.: 1, 16). This is mainly because researchers face a range of capacity constraints and barriers, at both individual and institutional levels. Barriers include 'inadequate KT knowledge and skills, particularly for communicating research and interacting with research end-users, insufficient funding, and inadequate institutional guidelines, structures and incentives promoting KT practice' (ibid.: 12).
In addition, it has also been shown that researchers undertake a narrow range of easier, less interactive KT activities, while investing little time to interact with different target audiences and to tailor their communication of findings to these audiences. The most frequent dissemination formats reported were scientific publications and conferences, whereas media were the least frequent avenue used (Murunga et al. 2020). Researchers engage more in 'producer push' than in 'user pull' or in exchange activities (ibid.). Producer push activities are described as research products in language appropriate to specific audiences, messages that set out specific actions, and attending workshops, conferences and other events (ibid.). In contrast, only a limited number of researchers engage in interactions outside their research process, such as participating in expert committees or groups, or in government-sponsored meetings (ibid.).
Researchers seldom engage in activities such as publishing non-scholarly articles, accepting requests from journalists to participate in interviews or debates, or making their research publicly available on websites, indicating that KT related to user pull was rare (ibid.). Furthermore, it is rare for researchers to align their research with government policy priorities, to actively interact with policy actors, and to synthesise evidence for policy and practice. Alignment with policy priorities was found more often in the area of health research than in articles describing government policy priorities and policy dialogues (Cheung et al. 2011, cited in Murunga et al. 2020.
In general, the most frequently mentioned outputs from research are single basic science and observational studies. Systematic reviews or other forms of syntheses are rarely mentioned, indicating that there is a lack of synthesising the evidence base to inform decisions on policy and practice (Cheung et al. 2011, cited in Murunga et al. 2020Uneke et al. 2012, cited in Murunga et al. 2020.
In contrast, in a small-scale study on reproductive health research in Bangladesh, researchers self-reported using multiple KT activities to share their findings with policymakers, although they did not seem to systematically track and assess their impact (Walugembe et al. 2015, cited in Murunga et al. 2020. However, only a small number of respondents developed policy briefs, had one-on-one meetings with policymakers or provided technical assistance to them, produced research reports, and engaged the media. Moreover, few respondents knew how and when during the policymaking process their findings were utilised, suggesting that they do not systematically assess the impact of their KT activities (Walugembe et al. 2015, cited in Murunga et al. 2020.
Another small-scale study included in the systematic review found that there are fewer and less diverse interactions between academic researchers and industry, including for products or services based on researchers' work, compared to interactions among academic researchers (Lashari et al. 2017, cited in Murunga et al. 2020. The situation may be changing, however. One Spanish-language systematic review of English-language academic publications investigated knowledge transfer by higher education institutions and found publications mostly about the global North, with only a few discussing the global South. The review findings confirmed that knowledge transfer can take a variety of forms, from commercialisation, academic or spin-off companies, patents or licences, to technology transfer or industry-academia relations . Higher education institutions have been increasingly interested in knowledge transfer, however, and have been integrating it into their strategies for innovation and social appropriation of knowledge. Strengthening relationships of trust between academia, businesses and state is significant for increasing the impact of knowledge transfer in wider society, through new products, services and solutions that go on to foster ecosystems of innovation (Guerra et al. 2018: 316).
Over the past decade there has also been a strong movement, particularly in health in Africa and South Asia, towards embedding science advice and evidence use in government ministries (Stewart 2015). Initiatives range from rapid response services ) to establishing dedicated multidisciplinary teams designed to break down barriers between the producers and users of research (Sen et al. 2017). Overall, it can be concluded that there is great variation in the degree and nature of interactions specifically between researchers and policymakers, depending on context and background.

Gaps in interventions conducted to improve KT capacities and practices
A number of past and current interventions and tools to improve KT by researchers and research institutions in LMICs have taken place, though very few have given rise to assessments published in academic peer-reviewed journals (Murunga et al. 2020: 17). These interventions have largely focused on improving researchers' individual capacity for KT through trainings and capacity-building -on KT (theory, application, barriers, and facilitators), research communication, systematic reviews, or developing and sustaining relations with policymakers and the media (ibid.: 17).
Other interventions have aimed to strengthen collaboration and to support the establishment and operation of networks for KT, such as EVIP-Net and the Consortium for Health Policy and Systems Analysis in Africa (ibid.: 17). Importantly, there has been a dearth of multipronged interventions aiming to enhance individual and institutional capacity for KT by researchers and research institutions in LMICs, according to Murunga et al. (2020: 17). This is despite evidence showing the crucial need for such approaches (ibid.: 17).

Actors specialised in 'knowledge management for development'
While knowledge management has a long history, some authors (Cummings et al. 2019: 3) argue that knowledge management for development is a more recent occurrence generated by large aid donors and actors in the global North for application in the global South. 12 Such KT remains primarily oriented towards the priorities and perspectives of these donors and actors (ibid.: 3). 12 The present review could not examine this claim due to time constraints. However, it may be worth noting that historians of Western and non-Western development and humanitarianism have documented attempts at knowledge management and KT that are much older than the 1990s, be it under conditions A 2017 consultation of experts suggested that UN-related knowledge brokering simultaneously displayed aspects of the third generation of knowledge management (inter-organisational communities of practice), fourth generation (knowledge processes embedded in organisational processes, and practicebased knowledge management), and fifth generation (local knowledge, and multi-stakeholder processes) (ibid.: 8).
This may be happening because the expertise exists at national level in LMICs, but also because financing for operations has been falling. Internal knowledge management and implementation in the field contribute to UN organisations' legitimacy as knowledge brokers, since the UN system has potentially hierarchical relations with other organisations working on development knowledge. The Sustainable Development Goals (SDGs) provide a helpful framework which both UN actors and others involved in knowledge management have referred to (ibid.: 5-6). Nonetheless, a specific framework on knowledge could be a useful complement to fill gaps in the SDGs on 'knowledge for development '. 13 Inter-organisational communities of practice, which are part of third-generation knowledge management, have played an important role in connecting professionals of knowledge brokering across and between organisations. Internally, many development organisations are concerned about losing the expertise of their employees who leave, and of consultants who work for them only temporarily. Thanks to knowledge communities with stable memberships, such as the IDRC-founded 'Knowledge Management for Development' (KM4Dev), expertise is not lost even as individuals move between organisations and consultancies (ibid.: 6).
Multi-stakeholder processes are important to facilitate links between the UN and local actors for knowledge management for development. Major facilitators of links include NGOs, social entrepreneurs, and the private sector in all its diversity, from multinational corporations to local small businesses (ibid.: 6). 14 However, local knowledge is inadequately covered, and thus brokered, by UN knowledge management. Some factors behind this are readily identified by UN experts themselves. For example, differences in languages prevent sharing of colonialism, imperialism, or greater local autonomy or independence. Lessons from those earlier attempts could well be useful, both to recognise historical legacies in today's KT, and to learn from successes and failures in a comparative, context-informed way. See, for example, Bámaca-López  knowledge between UN country offices and local people. 15 However, UN experts, such as those consulted in this study, tend to ignore more structural factors causing international development actors to marginalise local knowledge, and to sometimes drown out local voices. For instance, inequalities in symbolic power and convening power lead to hierarchies of credibility, where international organisations are at the top, followed by national governments, and local communities coming in a distant third (ibid.: 7).

Knowledge brokers active in the development field
The literature describes knowledge brokering in the development field, i.e. making connections between people to facilitate the use of knowledge, as complex, diverse, and contextual (Conklin et al. 2013, cited in Cummings et al. 2019Kislov et al. 2017, cited in Cummings et al. 2019. 16 Beyond these general observations, however, there are two major types of brokering, which have noteworthy differences. First, a widely researched type of brokering in development is about facilitating the use of knowledge from research in policy and practice. Such research has used varying terminologies, such as 'knowledge intermediaries', KT, and 'knowledge co-creation'. There are diverse models for such brokering, and many people engage in knowledge brokering without labelling their activities as such However, in development, people can broker knowledge at 'very many different levels, not just between research, policy and practice' (Cummings et al. 2019: 4). One perspective to understand this is to think of knowledge cultures. Each knowledge culture has its own types of content, inquiries and languages, and each individual is part of different knowledge cultures (Brown 2008, 2011, cited in Cummings et al. 2019. 15 The source article describes local languages as the barrier, but it could just as well be said that many UN officials' lack of mastery of local languages is the barrier. 16 Some practitioners in international development -mostly within the UN system -defined knowledge brokering by its role in adding value and by its functions: 'adapting, translating, connecting, acting as an intermediary, match-making, convening of networks and professional learning, connecting supply and demand for knowledge, catalysing and facilitating' (Cummings et al. 2019: 5). In this reference, 'development' is primarily understood in reference to the UN's SDGs, and refers to efforts to advance economic, social and environmental development, whether the effort is by local, national or international actors. Brokering thus occurs at many levels and in many ways. For example, 'brokers at the grass roots make linkages between these different types of knowledge' (Cummings et al. 2019: 4). Other development actors make linkages between local knowledge, scientific knowledge (e.g. extension services that aim to improve farming practices), and other knowledge from diverse sectors, such as health and agriculture.
Brokering therefore has a crucial role in development 'as a cognitive bridge between these different types of knowledge' (Cummings et al. 2019: 4).

Geography
Results presented in Annexe 2 show that all major geographic regions in the global South are covered. Different studies cover KT in different areas: some across the entire global South (Annexe 2), others across one or more regions, countries or subnational locations within in. The more limited the coverage, the more uncertain the generalisability and applicability of the findings about KT in other settings. However, studies with narrower geographic scopes can provide invaluable depth to findings, where wider scopes are strong on breadth.
There are differences in the number of references from the different regions. Most references are from sub-Saharan Africa (n = 30), followed by Latin America and the Caribbean (n = 15). Some areas do appear to be under-researched such as Western Asia, Middle East and North Africa (n = 2), and Asia and the Pacific (n = 8) (Annexe 2). Specific sub-regions within world regions seem underresearched too, such as regions in sub-Saharan Africa where French and Portuguese are among the official languages (such as parts of Western and Central Africa), compared to those where English is among the official languages (such as parts of East and Southern Africa). Besides, there seems to be some under-research on KT in rural areas, though this may be caused by the limitations in the keywords, as these excluded terms applicable to KT in sectors predominant in rural areas (for example, keywords about extension services for agriculture).
The knowledge base covers very diverse political, economic, social, and cultural contexts of countries overall. This being said, research seems scarcer on contexts marked by autocracy, by war or other forms of widespread violence (outside of KT in the international humanitarian field), by less formal or less powerful public institutions, and by economies that perform very poorly for development. It has been shown that country income status, country context, institutional culture, research type, and research topic all influence capacity for KT, with differences between and within countries explaining a multitude of variations (Murunga et al. 2020: 8-9, 16) while different sectors and academic disciplines are not associated with significant differences in KT strategies (Oliver et al. 2021: 62). Murunga et al.'s (2020) review found that KT capacities, factors, and interventions are largely consistent with what has been reported in high-income countries. However, there are differences between countries due to differences or inequalities in the factors that shape KT practices (ibid.: 18). For example, low-income countries have fewer KT capacities than middle-income countries, especially upper-middle-income ones (Maleki et al. 2014, cited in Murunga et al. 2020. There are also differences in knowledge about country-specific research priorities between countries (El-Jardali et al. 2018, cited in Murunga et al. 2020. Institutional culture is an important factor influencing KT capacity within and between countries (Ayah et al. 2014, cited in Murunga et al. 2020. For example, being part of an effective research network may be one specific positive factor (Gonzalez-Block andMills 2003, cited in Murunga et al. 2020: 8;Shroff et al. 2017, cited in Murunga et al. 2020Georgalakis et al. 2018).

Differences by research type and research topic
The coverage of sectors is unequal, but the specifics are hard to ascertain due to the use of different terms in different sectors to discuss KT. One third (n = 69) of all references concerned KT in a health context (Annexe 2). Other sectors such as education, agriculture, and topics related to the environment are represented as well. However, it is difficult to ascertain how much research there has been because of a variation in terminology to discuss KT, which may have resulted in an oversight of publications due words and expressions not included in the search strategy.
The overreporting of heath-related publications is a serious problem given the importance of multiple other sectors and of cross-sectoral work for development. Under-reported sectors include social protection; work and economics; housing and shelter; water, sanitation, and hygiene (WASH); land; agriculture; food; territorial planning (urban and rural); infrastructure; energy; ICTs; environment (including climate, and environment-related disasters); and security.
This pattern, combined with other aspects of the knowledge base, has negative consequences for how KT in the global South is studied and understood. For example, one well-documented hurdle in KT has been KT actors' dearth of attention to power and the fundamentally political nature of KT, from micro to ids.ac.uk Working Paper Volume 2023 Number 592 Knowledge Translation in the Global South: An Exploratory Mapping of the Literature 47 macro scales (for a discussion on this, see Oliver et al. 2021: 55-60). The failure to bring political science and economy into many KT efforts may well be one reason for this hurdle. Murunga et al. (2020) state that the type and topic of research influences the extent to which researchers are involved in KT activities (Nedjat et al. 2008, cited in Murunga et al. 2020Lavis et al. 2010, cited in Murunga et al. 2020. For example, health systems researchers were more likely than basic and clinical researchers to undertake KT activities targeting non-scientific audiences (Nedjat et al. 2008, cited in Murunga et al. 2020. In contrast, Oliver et al. (2021) state that policy sectors or academic disciplines do not determine the approach that researchers and decision makers choose in order to engage stakeholders in their work. Stakeholder engagement shares many features across sectors and disciplines. However, decision makers' and researchers' choices depend on a balance of interest between generalisable evidence and local evidence, and on how much consensus there is about what is known when collaborative work begins (Oliver et al. 2021: 62).

Differences in underlying theories of evidence-informed policy and practice
The use of explicit conceptualisations and theory by researchers and research institutions to design KT activities is limited (Murunga et al. 2020). However, a number of practitioner and policy KT efforts seem to have been based on explicit theories of change that spell out their understanding of evidence-informed policy and practice (for examples of discussions of interventions, see Devos et al. 2019;Yearwood 2018).
One point of agreement in the literature is that there is a diversity of concepts (and associated terminology) used to describe KT in the global Southsometimes but not always in reference to terms, concepts, and definitions from the global North, especially Canadian institutions' definitions of KT (for example, Hanlin et al. 2021;Kraemer-Mbula et al. 2020;Murunga et al. 2020;Oliver et al. 2021;Shaxson et al. 2012;UNESCO 2017). In the absence of explicit theories, concepts and terminology may therefore offer some entry points to identify tacit understandings by actors involved in KT about what KT is, how it works, and how evidence relates to policy and practice. Still, whether the entry point for analysis is theories or concepts and terminology, dealing with entirely or mostly implicit approaches poses serious challenges for secondary research conducted in a short time frame.

Challenges posed by implicit approaches to theories
Further, while explicit theories of KT are sometimes scant, there are always at least implicit theories embedded in interventions, but these are complex and time-consuming to ascertain through secondary research only. This makes it difficult to run comparisons and, for instance, to learn about the validity, effects, and relative impact of different theories. Among others, the following study challenges arise. KT actors, as any social actor, have at least implicit representations. So, whether KT actors spell out any theories or not, their actions will reflect their understanding of their situation and their goal, based on their ideas, values, and interests. However, this is harder to analyse than at least partly explicit representations.
Second, some KT actors may develop their explicit theory entirely inductively and during their action, so that they have no explicit theory at the start but do have one in later stages of their KT action. This means that any publication about KT action that this actor generates early on will be more complex to parse for the underlying theory in a review of secondary literature.
Third, there can be a divergence between a KT actor's stated theory as laid out in publications, and the theory their eventual actions reflects. This may be due to external constraints forcing them into a suboptimal course, preventing them from taking actions that would have matched their understanding of evidenceinformed policy or practice. Furthermore, circumstances may have made actors change course, or because they did not enact their stated intent (for example, actors may claim to adopt an inclusive approach, but end up excluding certain groups, wittingly or not) which results in challenges for secondary research.
Fourth, some of the studies found in the research (such as Cummings et al. 2019;Olivier de Sardan 2021) suggest that individual and collective KT actors may simultaneously harbour multiple, at times contradictory, theories about evidence-informed policy and practice. This can happen more so when the theories remain implicit, or when several actors with different theories are trying to collaborate on a single KT initiative. Therefore, any investigation should examine differences in theories both between the different modes of KT, and also within each KT mode.
Fifth, some KT endeavours are based on theorising that is deeply grounded in the context of intervention (akin to advocacy plans tailored to an issue and a particular context). In such a case, some of the crucial aspects of the theory would not easily be generalisable or comparable with other contexts or would only be so after in-depth analysis identifies the contextual conditions of history, society, and economy and how it shapes what constitutes 'evidence', 'policy' and 'practice', and clarifies the links and relative weight between generalisable variables (such as evidence and intended outcomes), and contextual ones.

Tentative findings as starting points for future analysis
Overall, the four modes of KT (Figure 1.1) and associated theories in the research framework match the actual KT practices documented in the literature well -in particular, the many activities that are supply-driven or demand-driven.
At the same time, the literature confirms that KT actors have multiple alternative or complementary ways to theorise KT and evidence-informed policy and practice, and that centring analysis on the axis of linear vs interactive approaches is useful, but not the only insightful approach. For example, the approaches and underlying concepts of knowledge documented in the expert study on generations of KT yields useful insights about how knowledge and its use are understood theoretically (see Cummings et al. 2019).
One important nuance the literature brings, compared to the hypothesised modes of KT, is that parts of the KT practices identified in the literatureespecially those that combine supply-and demand-side action -are not about formalised initiatives with a plan and theory (whether implicit or explicit), but about a myriad of informal activities and associated theories that develop in an organic way, probably in response to learning, opportunities, and constraints that arise over time. This covers, for example, what the synthesis of expert views calls KT as a 'cognitive bridge' between different knowledge cultures, especially as part of the fifth generation of KT that revolves around a system or ecology of development knowledge at society-wide level (Cummings et al. 2019).
In those cases, actors carrying out KT adapt their actions and pull different levers of KT activities as their understanding (and theory or theories) of how evidence could inform a targeted policy or practice in context evolves (Oliver et al. 2021).
Lastly, there remains an open question about interpreting the weaknesses and gaps in KT practices identified earlier in this paper, and how much of these shortcomings result from theories vs other factors. For example, researchers often use less interactive types of activities and do not diversify their target audiences much (Murunga et al. 2020). Do these shortcomings result from constraints on actions that KT would like to take but cannot? Do they result from KT actors' deliberate or unconscious theory of KT and evidence-informed policy or practice? Or do they result from a mix of both constraint and theory? The source material identified constraints, but there was not enough information on whether researchers' representations of KT also played a decisive role. This illustrates that further secondary research, and likely further primary research, is needed if comparisons of theories within and between different modes of KT are needed.

Drawing lessons from failures, successes, interventions, and contextualisation
The present research contributes to filling a gap in existing research on KT in the global South. Prior literature on KT has established that merely having knowledge about evidence into use processes is not enough to succeed at influencing policy and practice. In addition, it is unclear and contested how applicable knowledge about the global North is to the global South. Lastly, available secondary literature on KT in the global South has some shortcomings, such as a frequent focus on the health sector and on sub-Saharan Africa. By exploring recent knowledge on the global South in three languages across world regions, sectors, and academic disciplines, the present review offers a crosscutting view that identifies areas and topics that often get missed in similar reviews.
While this review is broad in scope and gives an indication of the state of the literature, it has several limitations due to time constraints and feasibility considerations. In particular, the results presented here are not a systematic review, and had to leave aside relevant areas of the literature that are associated with search terms not closely associated with particular expressions such as 'knowledge translation'. Whereas the reader can learn about gaps in this KT literature, this review cannot provide a comprehensive picture of current practices and innovations in KT across sectors and geographies. There are undoubtedly significant innovations underway in KT in the global South which have not yet been sufficiently documented in ways that would have allowed this review to pick them up.

Strengths and weaknesses in the state of knowledge
Despite its limitations, this review was able to identify and map a large number of relevant publications, thus offering a good vantage point on the state of knowledge. It confirms that there is a large knowledge base about KT in the global South. Overall, there is a high degree of methodological rigour, and a frequent -though not systematic -acknowledgement of the ethical, political, and epistemological issues involved. The knowledge base encompasses a good balance of academic, practitioner, and policy sources, and diversity in focus as references use their material to address KT through a more conceptual, theoretical, or empirical lens. The study approaches (e.g. single-or multi-case studies), types of empirical material used, and methods are also diverse and well balanced in the overall base. Interestingly, there seem to be differences in quantity and contents by language in publications, with references in French and Spanish, for example, discussing KT from generally more critical perspectives embedded in social sciences.
The knowledge base, taken in the aggregate, generally provides good coverage on many aspects of KT, even as the scope of individual studies is extremely variable and as some areas remain under-researched (e.g. Asia and the Pacific, rural areas, contexts of war or autocracy, longer study time frames). In particular, the knowledge base amounts to a large coverage regarding: the aggregate size of studied populations, groups or samples; geographies; types of political, economic, social and cultural contexts; time frames. There is also good though differentiated coverage about the main areas of change in processes and outcomes that KT may aim for (namely policy, practices, and collective action), about the causes, dynamics, and effects of both KT and the change that KT may aim for, as well as about the actors supporting KT, the actors targeted for change, and the types of knowledge being 'translated'.
The findings presented in the literature also seem to have multiple strengths. The knowledge base provides a balanced mix of descriptive, narrative, analytical, and synthetic findings. Many studies offer findings that are conclusive, not just indicative. Moreover, among the many references that examine some form of inference, a majority seem to establish causality, not just correlation. Lastly, highlevel findings about KT in the global South seem generally consistent across the literature. However, differences, nuances, contradictions, and oppositions do exist, especially in the less high-level findings. This is likely due to a combination of differences in study contexts, in study scopes and methodologies, and in authors' values, ideas, and interests.
A few weaknesses are present in the knowledge base and call for attention or caution. For instance, there is a recurrence of some specialised authors, editors, publishers, funders and commissioning organisations among the sources found. While this means KT has become a field with in-depth, long-term specialist knowledge, this also carries the potential risk of group thinking and of particular actors' outsized influence over the study and understanding of KT.
Another weakness is that the coverage of sectors is uneven and patchy: health and, to a lesser extent, education appear to be much more studied, while literature searches run for this research found barely any references that would focus on other vital sectors such as social protection, housing, or energy-related topics. Relatedly, there is coverage by all scientific disciplines, but medical sciences are over-represented, and some (

Strategies and variations in KT in the global South
Drawing on a few high-level literature reviews that have both cross-regional and cross-sectoral scopes, the paper synthetises information about KT in the global South by answering the four lead questions.

Research question 1: What are the primary strategies used to enhance the use of knowledge in Southern contexts?
The analysis draws on an overview of the types of 'knowledge management for development' identified through an initiative that brought together researchers in knowledge management outside and within the development field. This initiative identified five generations of knowledge management for development, each with different perspectives, methods, and tools (Cummings et al. 2019). Furthermore, types of knowledge and inequalities were also discussed.
The study concluded that, when the diversity of knowledge management is taken into account -from informal grass-roots work to formal multilateralism -current knowledge management for development brings together practice-based and societal approaches. Practice-based knowledge processes are embedded within organisations (for example, through the use of peer assistance or knowledge scans), and between organisations (for example, through best practices or peer assistance in inter-organisation communities of practice). As for societal knowledge management, it revolves around the integration and co-creation of knowledge across domains, and the ecology and ecosystems of knowledge. It includes multiple knowledges, with an emphasis on local knowledge, and on the importance of emergence and complexity in knowledge. It involves multistakeholder processes, including citizens and the private sector. It is oriented towards the global public good (as laid out, for example, in the UN's SDGs), and knowledge commons (Cummings et al. 2019).

Research question 2: How do these primary strategies differ in relation to key stakeholders and intended outcomes (changes)?
The review shows that researchers and research institutions practice KT only to a limited extent. A number of researchers in LMICs practice KT, but a significant part do not. Moreover, researchers resort to a narrow range of easier, less interactive KT activities, while investing little time to interact with different target audiences' needs and responses. They engage more frequently in 'producer push' than in 'user pull' or in exchange activities. In general, researchers interact comparatively more with policymakers, and less with industry, media, other stakeholders, and the wider public (Murunga et al. 2020).
The degree and nature of interactions specifically between researchers and policymakers seems to vary by context. Overall though, it is rare for researchers to align their research with government policy priorities, to actively interact with policy actors, and to synthesise evidence for policy and practice (Murunga et al. 2020).
Even when researchers did engage with target audiences, they did not seem to systematically track and assess their impact. More broadly, interventions to improve KT have largely focused on improved researchers' individual capacity for KT, and neglected interventions through institutional capacity-building and change, and through multipronged interventions to enhance both individual and institutional capacity (Murunga et al. 2020).
Looking at a different set of stakeholders, the previously mentioned study about actors specialised in 'knowledge management for development' concluded that this type of knowledge management has remained primarily oriented towards the priorities and perspectives of the large aid donors and actors in the global North that supported this field for application in the global South. However, this may be slowly changing, as other actors involved have been doing effective knowledge brokering. For example, the UN system and some other aid actors are turning more into brokers of internal and external knowledge than producers of knowledge, through inter-organisational communities of practice and multistakeholder processes. While international organisations often neglect and inadequately broker local knowledge, other actors such as NGOs are better at this (Cummings et al. 2019).
Meanwhile, there is an emerging literature on the movement in Africa and South Asia towards embedding knowledge brokerage functions in government agencies or creating dedicated client services in universities to meet the evidence demands of policy actors. These developments, although predominantly in health, may become very significant for how knowledge translation is operationalised and understood in a Southern context (Mijumbi-Deve et al. 2022;Sen et al. 2017).
Knowledge brokers active in the development field are a wide and diverse category of stakeholders, with two main subtypes that show noteworthy differences with one another. A first type aims to facilitate the use of knowledge from research into policy and practice, through processes such as knowledge intermediation, translation, or co-creation. This commonly entails building relationships and networks, and trust between participants, with brokers who are positioned as credible experts (rather than as advocates or lobbyists) and familiar with the domain and its stakeholders (Cummings et al. 2019).
The second, less well-researched type, is about providing a cognitive bridge between different types of knowledge. This relates to the many forms and levels of brokering that occur between different knowledge cultures. Each knowledge culture has its own types of content, inquiries, and languages (Brown 2008, 2011, cited in Cummings et al. 2019. For example, development actors can make linkages between local knowledge, scientific knowledge, and other sectoral knowledge (Cummings et al. 2019).

Research question 3: How do different strategies [used to enhance the use of knowledge in Southern contexts] relate to sectors, disciplines, and geographies?
At the widest geographic scale, findings about KT capacities, factors and interventions are largely consistent across high-, medium-and low-income countries. However, due to differences or inequalities in the factors that shape KT practices, KT capacity seems lower in low-income countries than in middleincome countries, especially upper-middle-income ones (Murunga et al. 2020).
Beyond this, differences between and within countries matter, separately and together. Country-specific factors are at play, though specifics on this are scarce in the reviewed literature. However, while differences between countries matter, they alone cannot explain variations in KT capacities: differences within countries play a major role too (ibid.).
An important source of variation, within and between countries, is institutional culture, though specifics on this are scant. For example, being part of an effective research network may be one distinguishing positive factor, in areas such as institutional incentives for KT, capacity-building for KT, and impact assessment of KT efforts -the indicative example in this case is the cross-regional Alliance for Health Policy and Systems Research (ibid.).
Even within the same institution, the type of research conducted is another important factor. For example, one study on the Tehran University of Medical Sciences in Iran found that researchers working on health systems were likelier to do KT than those doing basic or clinical research (ibid.).
Similarly, the topic of research, and the associated research culture, may influence researchers' KT. For example, in a large-scale, cross-regional study that examines the support provided by health research institutions to their researchers' KT, researchers working on childhood diarrhoea were consistently more likely to have worse KT capacity than researchers working on malaria prevention, contraception, or tuberculosis treatment, yet also more likely 'to believe that their research was ready for application ' (Lavis et al. 2010, cited in Murunga et al. 2020. In contrast, policy sectors and academic disciplines are seemingly not associated with significant differences in KT strategies adopted by researchers or decision makers to engage stakeholders in their work. If anything, there are many shared features across sectors and disciplines (Oliver et al. 2021).

Research question 4: How do the different modes of KT interventions [used to enhance the use of knowledge in Southern contexts] reflect different theories of evidence-informed policy and practice?
Only some KT interventions (and evaluations) documented in the literature appear to have been deliberately and explicitly based on any theory of evidence-informed policy and practice, be it at the outset of efforts or by the time of an assessment or the end of an initiative. This shortcoming may be particularly pronounced in KT done by researchers and research institutions. In the absence of explicit theories, concepts and terminology may therefore offer some entry points to identify tacit understandings. By contrast, while explicit theories of KT are sometimes scant, there are always at least implicit theories embedded in interventions, but several challenges make them complex and time-consuming to ascertain through secondary research only.
Due to this, the present research can only offer the following tentative findings as starting points for future analysis. Overall, the four modes of KT and associated theories in the framework for this research match the actual KT practices documented in the literature well. Nonetheless, KT actors have multiple alternative or complementary ways to theorise KT and evidence-informed policy and practice.
One important nuance the literature brings compared to the hypothesised modes of KT is that parts of the KT practices identified in the literature -especially those that combine supply-and demand-side action -are not about formalised initiatives with a plan and theory (whether implicit or explicit), but about a myriad of informal activities and associated theories that develop in an organic way, probably in response to learning, opportunities, and constraints that arise over time. Lastly, there remains an open question about interpreting the weaknesses and gaps in KT practices identified earlier in this paper, and how much of these shortcomings result from theories versus other factors such as external constraints on KT actions.

Lessons from failures, successes, interventions, and contextualisation
There are well-established factors that shape KT as enabling conditions or as obstacles. In broad terms, four types of factors are most influential. The first relates to the context for KT research use -namely, whether research gets used by target audiences to inform decisions on policy or practice is crucial. Power in the situations, structures and systems, such as political context, donor influence, and international influence, are major determinants (Murunga et al. 2020).
Hierarchies of knowledge about development, where local knowledge is devalued, is an obstacle (Cummings et al. 2019).
The second type of factor relates to knowledge producers. Factors here include research institutions' incentives and support for KT, as well as researchers' interest and skills in KT, and their understanding of context. A crucial skill is the ability to tailor, target, and time research communication for different audiences, and to choose the right messengers and channels. Factors also include researchers' credibility, reputation, and institution, as well as the credibility, relevance and timeliness of evidence, all as perceived by target audiences (Murunga et al. 2020).
The third type relates to knowledge users. One is decision makers' understanding of research. Another is the nature of the policy issues -including whether the issues are seen as technical, contested, or in the policymakers' interest (ibid.).
The fourth relates to the relations between knowledge producers and users, especially the relationships, incentives, and stances between decision makers and knowledge-producing individuals and institutions. In particular, both the quality and diversity of collaborations and networks with target audiences matters (ibid.).
As for interventions to improve knowledge producers' capacities in KT, an important finding is that multipronged interventions that improve individuals' and institutions' capacities are important for sustainable improvement (Murunga et al. 2020). Continuity of funding and long-term perspective are important too (Cummings et al. 2019).

Lessons from contextualisation efforts
The degree of clarity and consensus among stakeholders about (1) the evidence and (2) the context of decisions determines how researchers and policymakers can best engage stakeholders in context. This is because, in some contexts, generalisable evidence may not be available or clear, or not combined inclusively with context-specific evidence such as local data and tacit knowledge. Moreover, for successful KT, stakeholders must take into account the structural sociopolitical factors that shape how actors produce and use evidence, namely: power, interests, and politics; institutions, relationships, and alliances; ideas, cultures and world views (Oliver et al. 2021).
More broadly, in many contexts, the problems, factors and solutions in knowledge use are not primarily about knowledge translation, but rather about knowledge implementation (Olivier de Sardan 2021). Consequently, disciplines such as anthropology, sociology, history, and political science are invaluable when thinking about this. For example, Jean-Pierre Olivier de Sardan's research has looked at why so many national and international KT initiatives for development (such as participatory consultations, pilot projects, randomised control trials, trainings, etc.) have failed for decades to meaningfully change practices in Niger and other global South countries, in sectors such as healthcare (e.g. midwifery) and social protection (e.g. cash transfers). His anthropological research shows that local actors' 'practical norms' that stem from historical, economic, political, and social context are crucial to explaining the gap between intended KT plans and actual practices (ibid.).
Accordingly, the author concludes that fostering change towards better practices lies with empowering field-level 'contextual experts' as insider reformers with the skills to create change that starts from within local 'practical norms', social logics, and governance modes, rather than pushing 'travelling models' of best practices, which international and national development actors believe they have contextualised but which actually fail to account for frontline contexts of implementation. Internal reformers who are contextual experts have three sets of competencies: lived knowledge of the implementation environment from the inside; reflexivity, i.e. the ability to have perspective and critical distance towards the context; and a capacity for innovation (ibid.).

Critical reflections on the literature
This final section presents critical reflections about the state of knowledge and the contents of the literature reviewed, as points to consider: -Systematicity: think about KT in consistent frameworks, and reflect on gaps and disconnects. For example, recognise that some aspects of KT may not be documented and known on purpose, to protect some of the actors involved.
-Humility: know when KT is not the right question or the right answer. For example, recognise when local actors' 'practical norms' that stem from historical, economic, political, and social contexts are actually the key, rather than 'knowledge' as such (Olivier de Sardan 2021). • Think about what doing no harm means in KT, and apply it; • Re-connect knowledge on KT to other bodies of knowledge, and learn from those (such as political science on policymaking, or practitioner knowledge on advocacy); • Address the limitations of problem-solving, which dominates KT studies; • Beware of the limitations created by a focus on innovation; and • Decolonise what 'knowledge' refers to, doing so without turning the global South into a distinct 'other' from the global North.
-Justice: systematically build transformative, intersectional approaches to power into KT work.

Limitations
The present work has a number of notable limitations, some inherent in the research approach and topic, others related to issues in the search tools and knowledge base, and driven by the time frame of the project.
First, the core concepts in the research topic are contested, and have been described with a range of different terms. KT is a notoriously contested concept with varied definitions and uses (see, for example, the discussion in Graham et al. 2006). So is the use of the 'global South' terminology (see Box 1.2). This had two implications that generated limitations. One limitation is that the search syntax is bound to have missed possible relevant publications. In response to the diversity and contestation of terms about KT and the global South, combinations of search terms that would be as inclusive as possible to catch publications about the research topic were used. However, there may be some gaps remaining.
Second, unlike many prior in-depth or systematic reviews of KT which had a narrower search focus and -even then -did not claim to be exhaustive, the present research deliberately did not limit review by source type of study, sector, or geography (other than the entire global South). While this yielded rich results quantitatively and qualitatively, the large number of relevant results obtained led the researchers to narrow the scope in other ways to keep the results manageable for decisions on inclusion. One methodological adjustment made was to limit searches to publications from the past five years. This, of course, means that highly relevant references published before 2017 have been missed.
On top of the limitations generated by the research approach and topic, some limitations to findings and reproducibility stem from the search tools and knowledge base themselves. This is worth emphasising as these problems are likely to particularly hinder researchers from the global South, who often have fewer resources such as budget, support staff, and IT tools to tackle them. The problems are twofold: some of the relevant search tools and knowledge are not openly and freely accessible, and even with tools and knowledge that are, some are not easily accessible. For example, to search the academic literature, extensive use of the Web of Knowledge academic database was made, which has a lot of time-saving options to filter and refine searches compared to Google Scholar. However, access to this database is not free, and may be too costly for researchers outside academic institutions, as well as for many researchers in the global South, to access. Furthermore, a number of references -especially, but not only, academic ones -are behind paywalls. In several cases, the full text of some references was not accessible and, therefore, content may have been missed.
In sum, given the limitations of the present research, this review is exploratory, and does not claim to be a systematic review, meta-review or similar. It is not based on systematic searches, but rather on structured searches combined with non-systematic snowballing and on purposive recommendations. It is likewise not based on systematic content analysis, but rather offers a mapping of relevant publications, and an initial documentation and critical discussion of Southern KT models. There are undoubtedly significant innovations underway in KT in the global South which have not yet been sufficiently documented in ways that would have allowed this review to pick them up.
The limitations identified above have implications for the scope and findings, but also for the reproducibility of this research due to the non-systematic, partially purposive nature of the searches.

Summary
This analysis provided an exploratory review of recent academic, practitioner, and policy literature in English, French, and Spanish in order to document and discuss KT practices in the global South. Its thematic analysis used a combination of deductive and inductive methods and enabled a critical examination of the state of knowledge. The research then built its substantive findings based on a synthesis of contents and reflections about the knowledge base. The research was thus able to categorise the KT strategies used in the global South and to identify differences by stakeholders and goals, and by sectors, disciplines and geographies, before discussing the often-tacit theories of evidence-based policy and practice that underly KT. Furthermore, it identified lessons drawn from KT failures, successes, interventions, and contextualisation. This included identifying key factors shaping KT processes, outcomes, and interventions. It also included engaging with the nuances of contextualisation, starting from the observation of current shortcomings in KT and conditions of uncertainty about evidence and context, and building up to lessons and pathways to successful adaptation of KT to different conditions. The remainder of this conclusion will offer suggestions for further primary and secondary research on KT in the global South.

Suggestions for further secondary research on KT
Given the limitations of this research and the gaps in coverage in the literature found for this review, some additional secondary research would be valuable, though on targeted areas only to avoid duplicating work already completed by other literature reviews and syntheses. One constant that is worth keeping is the use of multiple languages, especially to avoid using only English. Indeed, one interesting complement to the present work could be to conduct searches in academic and grey literature in a few more languages -for example, German, Italian, or Mandarin (Chinese) -as there are a plethora of academic journals available that publish exclusively in these languages.
Other interesting lines of enquiry could include using additional search terms to elicit results from further sectors, disciplines, and thematic areas. Additional academic databases could also be used, such as Cairn for French-language literature. A final possibility would be to conduct secondary research on how KT is done in media, or on videos and audios produced by grass-roots actors.

Suggestions for further primary research
The systematic review of KT by researchers and research institutions in LMICs identified three key issues to understand researchers' KT capacity and practice and to identify effective interventions (Murunga et al. 2020: 14): 1. The need for more and better research on capacity and practice; 2. The need for multifaceted interventions that address gaps in capacity and practice, for both individual researchers and research institutions; and 3. The need for better designed studies that evaluate interventions.
There needs to be more and better research, and a stronger evidence base, about KT in LMICs, the authors of the systematic review conclude, adding that this need for more high-quality literature on KT exists in high-income countries too. A greater number of research and evaluation studies of higher quality is needed on all aspects and influencing factors of researchers' and research institutions' KT in LMICs: capacity; practice; and effective interventions to strengthen KT capacity (ibid.: 13, 16, 18). This should also include some standardisation of the methods and tools used to assess capacity and practice, in order 'to aid comparison across research type (basic vs applied), research topic, institutions and contexts' (ibid.: 16, 18).
In line with Murunga et al. (2020), the following recommendations can be made: -Extend beyond case studies and purely descriptive studies; -Base investigations on theory; -Draw on literature and methods of policy analysis from political science; -Use participant observation and documentary evidence; -Provide nuanced interpretations of 'context', 'policy', and 'research'; -Investigate researchers' interest in KT, and how this influences their KT practice; and -Pay attention to the range of impacts and effects of research.
To evaluate interventions in particular, KT practitioners and scholars are recommended to use 'realist approaches, pragmatic trials, impact evaluations, implementation research and participatory action research' (Murunga et al. 2020: 18). These methods are more suitable to evaluate social, contextually sensitive interventions. In addition, evaluations should pay greater attention to the variety of impacts and effects resulting from research (ibid.: 18). As a complement to the suggestions by Murunga et al. (2020: 18), there is a further need to include quantitative, qualitative, and mixed-method research that generates primary material, theories, identification of key issues and dynamics (and of key factors, if causalities can be determined) about: -Inequalities and intersectionality in KT; -Fine-grained contextualisation of KT informed by political, economic, social, and cultural considerations, based on social science approaches that take power into account; -Comparative case studies, with comparisons between both similar cases and between different cases; and -Most aspects mentioned under the earlier suggestions for additional secondary research, and related ones that lend themselves to primary research (for example, research into the use of art or storytelling).
Throughout, it would be valuable to learn from successes, failures, and average outcomes of KT efforts -all three defined broadly, so they relate not only to easily visible outcomes such as policy changes, but also to less visible effects such as changes in processes, relationships, norms, institutions, or in values, ideas, interests, and power.

Annexe 1: Structured bibliography
This annexe offers a mapping of publications through a structured bibliography. It presents publications that are cross-cutting, that cover more than one geographic region, and/or more than one sector. These are the widest-scope references the search resulted in. The annexe lays out the most relevant references found, grouped according to the prioritisation criteria laid out in the methodology section of this paper.