The momentum of gender and the SDGs: a corpus-assisted study of Norwegian development aid discourse

The Sustainable Development Goals (SDGs) constitute the most comprehensive and far-reaching example of a trend concerned with “global governance by goal-setting” (Biermann et al. 2022). The goals were adopted by the United Nations (UN) General Assembly as the core of Agenda 2030, a wide-ranging “plan of action for people, planet and prosperity” (United Nations General Assembly, 2015). There are 17 SDGs, specified through 169 associated targets whose realisation is monitored across >230 indicators (United Nations General Assembly, 2018). Whereas a goal is general in nature (e.g. achieve gender equality), a target outlines more specific concerns (e.g. violence against women and girls) and an indicator serves to assess to what degree certain targets have been achieved (e.g. number of victims of gender-based violence by age and place of occurrence). Our focus on gender is motivated by the significance of the concept for the sustainable development agenda writ large. Gender inequality, conventionally perceived as socioeconomic inequality between men and women, is recognised by the UN as a significant obstacle to development globally (United Nations, 2022). Furthermore, various forms of gender-based discrimination, oppression and violence persists around the world in ways that threaten the lives, well-being and prosperity of women, men, and anyone existing outside the established gender norms of their societies (Madrigal-Borloz, 2023).

Gender-related concerns recur across the Agenda, as well as in the specific focus of goal 5, which is aimed at “achieving gender equality and empowering all women and girls” (Odera and Mulusa, 2020; United Nations General Assembly, 2015). In the phrasing of the goal, we note the immediate shift from gender to women, which marks the tension, characteristic of international norm-setting, between different approaches to the notion of gender and its relation to biological sex. Historically, the move from fixed binary oppositions to a more complex view of social gender norms in development discourse is associated with the evolution from a WID (Women in Development) to a GAD (Gender and Development) perspective, with the latter attempting to bring a variety of “political-economic power relations” into clearer view (Krook and True, 2010, p. 115).

The SDGs are a central soft governance framework for development aid policy, planning and evaluation, and the purpose of this paper is to explore their discursive impact on gender-focused development aid planning by Norwegian development aid NGOs. Norway is an interesting case to explore due to the country’s longstanding commitment to gender equality as a priority for development policies and aid programmes (Seims, 2011, Østebø et al. 2013), as well as its image as a reliable aid donor and UN supporter. This research studies a custom-built corpus of funding applications from four Norwegian aid NGOs. Applying a temporal, lexical analysis of documents produced between 2010 and 2020, we explore the discursive impact of the adoption of the SDGs in 2015 on these aid programme plans. Specifically, we ask whether the SDGs have shifted the conceptual scope of gender, and the development issues gender is seen as pertinent to, in the texts under study. This is particularly interesting to explore when keeping in mind that the strong focus on gender in Norwegian aid policies predates the SDGs, while the SDGs in turn must be seen as a compromise between the progressive and conservative approaches to gender issues that exist in global development policy negotiations. The hypothesis we aim to test, based on the widely reported discursive impact of the SDGs, as discussed in the next section, is whether the introduction of the Agenda has had a linguistically salient transformative effect on the NGO discourse under study.

Our methods build on previous work in the field of corpus-assisted discourse studies (CADS), an approach that allows researchers to combine quantitative insights of corpus linguistics with qualitative examinations of a text’s discursive features (Gillings et al. 2023; Partington et al. 2013). The article is divided into four sections. First, we conduct a review of relevant literature on the concept of gender, its importance to the SDGs, and its contentious status in development aid discourse. Second, we discuss which organizations, and in particular which texts, were selected for study. We also explain in more depth how we identify and analyse linguistic patterns. In the third section, we present our numerical observations, as well as some extended examples of the discourse around gender found in the corpus of Norwegian development aid documents. In the final section, we highlight our contribution both in terms of method and in terms of our findings about the discursive relationship between human rights, the SDGs, and the textual representation of gender in the corpus.

The relationship between linguistic innovation and sustainable development is a relatively new domain of enquiry (McEntee-Atalianis and Tonkin, 2023). Within the scholarship engaging with the implementation and impact of the SDGs, few engage specifically with the discursive impact of the goals on development aid programmes. A notable exception is a wide-ranging review by Biermann et al. (2022), which explores the political imprint of the SDGs across global and national policy development, practice, and discourse. The review concludes that the impact of the SDGs “has been largely discursive, affecting the way actors understand and communicate about sustainable development” (Biermann et al. 2022, p 795). This echoes the argument by Beisheim et al. (2022, p. 28) which holds that while there is little evidence of profound institutional change, the introduction of “novel concepts and narratives” is widely attributed to the goals. These novel concepts and narratives include phrases detailing overarching principles such as “leave no-one behind”, as well as narrated links between seemingly disparate concerns such as human rights and environmental conservation (Beisheim et al. 2022). These prior works thus point out that currently, a central point of reference for an assessment of the transformative potential of the SDGs is their capacity to influence discursive realities.

In this paper, we are particularly interested in the influence of the SDGs on development aid programmes, and thus in the operationalisation of global norms and standards. Scholars working on norm diffusion, notably Moallemi et al. (2020, p 300), argue that any material impact of the SDGs will rely on processes of norm diffusion and operationalisation. Global aspirations need to connect with concrete local programmes and proposals, and take into account the specific ecological, social, and economic interests of particular “groups, organizations, and individuals” (Moallemi et al. 2020, p. 300). Furthermore, as is repeatedly highlighted by the UN, evaluating whether or not global progress has been made in relation to a particular goal depends upon the availability of “internationally comparable data for SDG monitoring” (United Nations, 2022). Hathie (2020) highlights the need for mutually informative processes of disaggregating relevant data and recognizing local interests as especially important with respect to the core ethos of the SDG agenda, which involves reaching “the most vulnerable” first in an effort to “leave no one behind.” Abualghaib et al. (2019) illustrate the complexities involved in identifying intersectional vulnerabilities, specifically in the case of people with disabilities. They argue that while this population is recognized by the UN as a vulnerable group, particular vulnerabilities will only gain visibility if intersecting factors such as gender, income, and ethnicity are considered in conjunction with specific modes of disability. These factors remain difficult to measure, as well as hard to define.

The literature’s repeated mention of problems of definition and categorization points towards the central concern of this paper: the relationship between the flexibility of the discursive unit “gender” and the all-encompassing claim of the SDGs’ pledge to “leave no-one behind” (United Nations, 2017). Some scholars, notably Elizabeth Mills in an early report from 2015, have argued that the SDGs’ pledge to leave no one behind must necessarily entail a commitment to the eradication of injustices and inequalities relating to sexual orientation, gender identity and expression, or SOGIE (Mills, 2015). Mills does not, however, engage with the term gender itself as a concept that is used to categorize, specify and operationalise priorities in SDG implementation. We argue that the use of the term gender is particularly important to engage with due to the tension between the multiplicity of meanings that the term is made to cover. In international settings, gender is invoked, often quasi-simultaneously, to address biological differences between women and men, sociocultural roles and relationships, as well as a broader spectrum of questions related to identity and expression.

Vandeskog et al. (2022) perform a critical concept analysis of gender in the text of Agenda 2030. They draw attention to the fact that the SDG targets strongly reinforce a gender binary, by means of formulations such as “End all forms of discrimination against all women and girls everywhere” (5.1, emphasis added) and “By 2030, ensure equal access for all women and men to affordable and quality technical, vocational and tertiary education, including university” (4.3, emphasis added). We draw in our analysis on the findings presented in this article, which makes use of the analytical vocabulary of Laclau and Mouffe (2001) to capture how the signifier gender functions, in Agenda 2030, as a nodal point that fulfils particular discursive functions through the articulation of chains of equivalence that establish specific associations: while concepts such as women, (in)equality, discrimination and violence figure in close proximity to gender, concepts such as orientation, identity, and expression, associated with the sexual orientation, gender identity and expression (SOGIE) discourse, are relegated to the fringes of UN discourse. In the analysis below, we examine the occurrence of such conceptual clusters in a large corpus of Norwegian development aid documents, in order to sharpen our view of the potential discursive impact of the SDGs on the discourse of Norwegian development aid organisations.

In this study, we employ a corpus-assisted approach because we aim to identify patterns, both static and shifting, across a large number of texts. Corpus-assisted discourse studies allow for an iterative move between aggregated findings and specific cases that is particularly useful when exploring nuances of meaning-making in a complex discursive setting. Scholars that have applied a corpus-assisted approach to the study of sustainability discourse include Parlakkaya and Stenka (2024), who focus on discursive features in corporate settings, and Hassan (2022), who examines the relationship between interpersonal modality and ideology in Agenda 2030. Brun-Mercer (2021) conducts a corpus analysis of debates at the United Nations and pays specific attention to the representation of women and men by international decision-makers, for instance in relation to mentions of gender-based violence. Beyond the realm of sustainability and development discourse, corpus-assisted methods have been used to analyse the linguistic representation of gender in a variety of media environments, such as advertisements, newspapers, and online magazines (Baker, 2014; Karimullah, 2020).

As gender terminology has become both more complex and more conspicuous in the discourse of international organizations, it has also evoked considerable suspicion and antipathy. Graff and Korolczuk (2022) describe how what can be perceived as a 21^st-century “global anti-gender movement” relates to a broader conservative backlash against aspects of the neo-liberal hegemony. Gender discourse is here seen as having a destabilizing effect that contributes to accelerating processes of individualism and alienation that seem ironically to characterize an increasingly interconnected and globalized world. Graff and Korolczuk (2022) further demonstrate how this anti-gender movement brings together the populist right with more traditionally conservative forces.

The discursive momentum gained by the concept of gender is thus precarious, prone to backlash, and involved in a highly politicized, continuous process of “translation,” which refers here not just to interlingual, but also to intralingual processes of conceptual adaptation and contestation (Pope et al. 2020). Thus, the concept of gender is subject to a high level of contention and ambiguity. At the same time, the concept is called upon to guide practical development efforts under the SDGs. Consequently, there is a need for closer examination of the conceptual roles gender is made to fulfil in the textual construction of actual development programmes. With this paper, we aim to contribute to the body of literature engaged with the complexities of the gender agenda in development policy and practice and bring this issue into conversation with scholarship on the impact of the SDGs on development discourse and priorities.

The field of corpus-assisted discourse studies is rooted in “the search for – and belief in the importance of – recurring patterns” (Partington et al. 2013, p. 8). Recurrent linguistic events, such as common combinations of words, are deemed indicative of how social reality is enacted and maintained, and linguistic patterning thus provides an informative perspective on the emergence and functioning of social roles and institutions (Stubbs, 2010). Quantitative means of analysis, including frequency counts and collocation measures, can add rigour to and broaden the scope of the range of patterns a researcher can identify, whereas close engagement with textual fragments can safeguard research from overly decontextualized observations (Baker, 2012, Cheng, 2012). The analysis presented in this article is fundamentally based on the observation of the frequent co-occurrence of lexical items, and thus depends on basic measures of collocation as well as on the examination of keyword-in-context (KWIC) concordance lines. A KWIC concordance is, essentially, an organized list of every instance of a specified word or other search string in a corpus, presented along with a given amount of surrounding text for each result (Sinclair, 2003, xiii). In this study, we focus on lexical patterns that emerge around the concept of gender in Norwegian development aid NGO documents. Those patterns indicate how certain identities and vulnerabilities are presented as visible and relevant (or not) to the aid efforts in question.

We have selected four Norwegian NGOs in this study for a number of reasons. Norway plays an important role in the shaping of global development aid policy. The country has cultivated an image of itself as committed to multilateralism, and as a strong supporter of the UN. Norway is a large aid donor per capita, and strongly prioritizes gender equality in its development policies and development aid programmes, which are in part implemented through the Norwegian development aid NGO sector (Seims, 2011; Østebø et al. 2013). Funding for aid programmes is, in large part, distributed through multi-year framework agreements established between the Norwegian agency for development cooperation (NORAD) and individual NGOs. In applications for these framework funding agreements, NGOs must formulate plans that respond to Norwegian development aid priorities, while also accentuating their own core values and focus areas. In response to the adoption of the SDGs in 2015, NORAD has explicitly established the expectation that funding applications should demonstrate how proposed activities will contribute to the realisation of the SDGs (NORAD, 2018). Within the Norwegian development aid sector, there is thus a clear expectation of discursive change in relation to Agenda 2030.

In this article, we seek to examine the extent and nature of this change in relation to the concept of gender by drawing on a custom-built corpus of 154 texts, amounting to ~900,000 tokens. All documents in the corpus, hereafter referred to as the Norwegian Development Aid Corpus (NDAC), were originally submitted to NORAD in English, as part of funding applications by four organisations: CARE Norge, Kirkens Nødhjelp (Norwegian Church Aid, NCA), Folkehjelpen (Norwegian People’s Aid, NPA) and Redd Barna (Save the Children Norway). These organisations were chosen based on their relative size within the ecosystem of Norwegian development aid NGOs combined with the fact that the majority of their documents are available in English. They are all consistently among the top-ten recipients of NORAD funds in the period in question, and they are significant actors in the Norwegian development policy debate, including on gender issues. It is important to note that some of the documents may also function in other national contexts, given that organisations such as Save the Children have a global presence and produce international policies. The NORAD archives are accessible via a web portal for public access to government correspondence archives (eInnsyn.no), and are searchable by correspondent, date, and file title. Documents included in the corpus were sourced via the online archives, with assistance from staff from NORAD and the respective NGOs. An overview of the included documents is provided in Table 1. The function of all the documents included in the corpus is twofold: they are intended to communicate to donors and partners about the organisation’s work, priorities, and values, and also to communicate internally to guide and streamline organisational activities and processes. Analysing these documents thus provides insight into the impact of the SDGs on the way in which “gender” is mobilised to construct beneficiaries of and priorities for Norwegian development aid.

While all documents are thus drawn from a single, coherent discursive environment, certain distinctions within the material are relevant to the analysis presented below. Firstly, the documents can be divided into policies, strategies, and programmes, which we will consider as ‘genres’ of development aid discourse. Policy documents present a general outline of an organisation’s principles and values. Strategic documents describe an organisation’s rationale for its priority areas and modes of intervention. Programme documents, finally, outline plans for specific development aid projects at the operational level. A second distinction, as mentioned above, is between documents produced before and after the 2015 introduction of the SDGs. When both axes of differentiation are taken into account, the corpus effectively comprises five interrelated subcorpora: strategies pre-2015 (11 documents), strategies post-2015 (12 documents), programmes pre-2015 (55 documents), programmes post-2015 (69 documents), and policies post-2015 (7 files). Our archival search did not produce any policy documents produced before 2015. These documents are not mandatory attachments to the proposals, and it is possible that including them has become more common in recent years. We could therefore not compare policy texts before and after 2015. As illustrated in the next section, the five subcorpora can be combined and compared in different ways to produce insights into the impact of the SDGs on Norwegian development aid discourse.

As noted, we aim to examine whether the adoption of the SDGs introduces substantial variation in usage patterns pertaining to gender, as texts produced by NGOs and other relevant development actors are assumed to draw on a novel vocabulary that must eventually inspire institutional change. Indeed, the literature on the impact of the SDGs suggests that they have had significant discursive effects. However, it remains unclear “how deep these discursive changes go,” and whether they constitute mere rhetorical nods, or a more profound normative reorientation (Beisheim et al. 2022, p. 40). Indeed, there is the obvious possibility that “organizational engagement” with the goals is limited to “SDG-washing,” meaning that the Agenda is selectively referred to only when it can be painlessly embedded in existing patterns of practice and discourse (Heras-Saizarbitoria et al. 2022). In this respect, assessing the nature and extent of discursive change is challenging, and before moving on to the analysis, the main methodological challenges that we encountered during the research process should be highlighted.

A first methodological complication is the fact that no two texts are completely identical, and that any discursive environment can be expected to display some degree of inherent variability. Even though the SDGs have a standard formulation, they are continually adapted to suit particular local contexts, and they aim to intervene in discursive environments, such as development aid programmes, that are themselves subject to continuous rearticulation. It is therefore not self-evident to separate the observation of meaningful discursive change from default processes of reformulation. The second methodological challenge mirrors the first: despite their continuous adaptation to different contexts and purposes, the standard, homogenous formulation of the goals is widely invoked in a variety of settings, including development aid documentation. The resulting problem, in the context of the present research, is as follows: CADS can usually avail of a concentric consideration of expanding patterns, with collocational observations eventually giving rise to the identification of extended “units of meaning,” which are then related to the social context a particular corpus is meant to reflect (Stubbs, 2009; Buts, 2020). If one were to follow a similar approach focused on the concept of gender in a corpus of texts influenced by the SDG’s, however, it is highly likely, as our data also bear out, that the most salient extended textual unit will be “achieving gender equality and empowering all women and girls,” which is simply an exact replica of SDG 5’s standardized phrasing. Indeed, it is not uncommon for the more recent texts in the NDAC to explicitly list which SDGs are addressed by particular programmes or guidelines, and often this happens in designated sections or texts boxes which fulfil a preliminary framing function but are only to a limited extent embedded in the surrounding discourse. While this framing function is important, and will be addressed in our analysis below, the notable presence of prefabricated discourse patterns around concepts such as gender can occlude, from a quantitative as well as qualitative perspective, other relevant, but less frequent patterns.

The challenges thus have to do both with the heterogeneity of the texts that we study, and with the homogeneity of textual patterns related to the SDGs. These challenges relate to two methodological questions that have only recently received due attention in corpus approaches to discourse, namely: how to divide up complex collections of texts into subsets to make meaningful diachronic observations (Marchi, 2018), and how to approach the study of discursive change without discounting indicators of discursive stability, continuity, and similarity (Taylor, 2018). Both issues remind us that discursive effects are always relative, and thus better assessed in comparative than absolute fashion. Therefore, in the analysis presented below, we first contrast language patterns around the concept of gender in Norwegian development aid discourse before and after the introduction of the SDGs. We then contrast observed patterns in strategy and programme documents. This allows us to test the hypothesis that the introduction of the Agenda has had a more linguistically salient transformative effect than the ordinary processes of adaptation observable across texts produced by NGOs for a variety of related purposes.

In our study, the notion of ‘language patterns’ refers, at the most basic level, to the phenomenon of collocation, or the statistically observable, habitual co-occurrence of particular lexical items within a circumscribed textual span (Sinclair, 2003). In the analysis, we use the AntConc corpus software (Anthony, 2022, version 4.1.4) to draw up lists of collocates for gender across the subcorpora that make up the NDAC. If the list of top collocates remains relatively stable across subcorpora, this is interpreted as a sign of discursive stability, whereas variation is interpreted in terms of discursive change. Observed change can then potentially be attributed to the main variable separating the subcorpora under consideration, which can either be temporal (pre- or post-SDGs) or genre-based (strategy or programme). To assess potential changes in more detail, we complement quantitative observation with qualitative analysis, based on the examination of concordance lines and larger stretches of text, as is generally characteristic of corpus-based critical discourse analysis (Nartey and Mwinlaaru, 2019). Indeed, collocational phenomena point towards the textual construction of larger phrasal representations that constrain the discursive value of particular concepts, and in this sense, collocations provide an important index of lexically embedded institutional and ideological value frameworks (Salama, 2011). We address this issue by not only drawing on tabular lists of collocations, but also on collocation networks, which can simultaneously highlight a variety of extended relationships between multiple lexical nodes (Brezina, 2018). For the identification and graphical display of collocation networks, we make use of the LancsBox software package, in particular its GraphColl functionality (Brezina et al. 2015, version 6.0). Since the output of the tool is also based on collocation, the investigation of networks can only be an effective countermeasure against the exaggerated influence of prefabricated patterning if additional concepts of interest are identified at the outset of the analysis.

At first encounter, our analysis suggests that the NDAC subcorpus consisting of all policy, strategy, and programme documents produced after 2015 is profoundly influenced by the introduction of the SDGs. AntConc’s frequency list functionality indicates that the abbreviation ‘SDG’ occurs in 74 out of 88 included documents and takes up the 86th position in the overall frequency list, thus appearing more than very common items such as ‘policy’, ‘between’, and ‘there’. A perusal of the relevant keyword-in-context (KWIC) concordance, which shows a limited stretch of co-text on both sides of a given search word, reveals which specific SDGs (in their most concise consensus formulation) are most frequently invoked in the Norwegian development aid documents under consideration (Table 2).

The table indicates, firstly, that a majority of SDGs is referred to on a regular basis. Only SDG 7 (renewable energy), SDG 9 (innovation and infrastructure), and SDG 14 (life below water), receive no specific mentions at all, indicating that environmental sustainability (‘planet’) is not a core priority for the NGOs represented in our corpus. Three goals stand out as being of exceptional importance, namely SDG 5 (gender equality), SDG 10 (reduced inequalities), and SDG 16 (peace and justice), indicating a strong concern for social sustainability (‘people’). The third pillar of Agenda 2030, namely economic sustainability (‘profit’), is not irrelevant to the NGOs discursive positioning, but seems to play second fiddle to the social component. The frequency count also confirms the centrality of the concept of gender in the corpus, which strengthens the rationale for the focus of our analysis.

As explained above, the abundant mention of gender in relation to the SDGs does not necessarily result in discursive changes beyond the context in which the goals are explicitly mentioned. AntConc’s plot tool, which shows the position where search words appear within particular documents, indicates that occurrences of SDG in the corpus are often highly concentrated in, and thus restricted to, a particular area. In most texts, the relevant section is introduced by a formulaic phrase such as ‘We present below an overview of the main objectives of this programme and which SDG they contribute to’. SDGs are then individually listed, after which the text resumes its explanation of, for instance, relevant national policies and local partners, but without further addressing the connection to the goals. To assess, then, whether 2015, the year in which the SDG agenda was adopted, inaugurates a significant discursive reorientation as regards the concept of gender, we firstly use AntConc to calculate the 30 top collocates (default likelihood settings, ranked according to MI3 effect size, minimum range and frequency of 5)^{Footnote 1} of gender in pre-2015 strategy and programme documents, and contrast the results to those for the post-2015 corresponding texts in the NDAC. The data is presented in Fig. 1. At this stage, we are primarily interested in the extent, rather than the nature of observed changes, which will be discussed later. Thus, as explained above, we provide an additional contrast between all strategy documents and all programme documents in the NDAC (Fig. 2, identical measures) in order to assess whether the discursive impact of the SDGs is greater than the impact of general textual adaptations to genre expectations. We excluded from all lists items such as and, the, and of, as well one-letter types such as c and e. We included 5 additional collocates at the bottom of each list to account for potential slight variations in ranking. Items which appear among the top collocates in only one of two contrasted lists are shaded orange, whereas stable items are shaded blue.

Top collocates for gender in the pre-2015 and post-2015 subcorpora.

Full size image

Top collocates for gender in the strategy and programme subcorpora.

Full size image

From a numerical point of view, the two tables behave in a remarkably consistent manner. Out of the 30 top collocates from texts before 2015, 13 do not occur in the list of 35 strongest collocates after the introduction of the SDGs, while 15 new items are introduced. Similarly, when contrasting strategy with programme documents, 14 items do not reappear, and 15 are added. While one could argue that a change of ~50% of the collocates seems to be strongly indicative of extensive change, the outcome of the comparison does not support the hypothesis that the discursive impact of the SDGs is greater than the effect of adaptation to genre expectations. Upon closer inspection, it also becomes clear that the inventory of items solely occurring on one side of both tables, for instance in the pre-2015 table and in the table representing strategy documents, run parallel to some extent. The collocate analysis, for example, disappears from the right-hand side of the table when contrasting based on temporal as well as genre distinctions. As both tables contrast the same set of documents, the cause of this parallelism is obvious: a subset of the data from the pre-2015 strategy documents displays distinctive collocational patterns.

In fact, a single document is responsible for the strong presence of the collocate analysis. The document is a toolkit that aims to explain the concept and facilitate the application of gender analysis, defined in the relevant text as the collection, analysis, and interpretation of “data and information about the specific situations, roles, responsibilities, needs, and opportunities of girls, boys, women, and men” (Save the Children, 2014). The toolkit has the specific function of making strategy guidelines more operational in programmes, and as such conceptually mediates across the genre boundary established in the table above. When the collocation analysis is repeated without this particular text, the differences between the list of collocates before and after 2015 becomes less pronounced, while the differences between strategies and programmes become more visible. In the NDAC, discursive variation can thus be said to be more dependent on the granularity of genre differentiation than on the temporal divide between documents produced before and after the introduction of the SDGs. Consequently, the development aid documents under consideration could presumably accommodate the language of the goals without having to fundamentally alter their conceptual repertoire.

The tentative observation of conceptual continuity can further be considered with reference to specific collocates. In the previous section, we noted that research on the standard formulation of the SDGs has highlighted the absence of sustained reflections on gender identity, orientation, and expression, in conjunction with the strong association of gender with women, (in)equality, discrimination, and violence, (Vandeskog et al. 2022). Figure 1 shows that all the latter concepts are indeed strongly associated with gender after the introduction of the SDGs, but also that this discursive embedding was already well-established before 2015. A slight increase in effect size scores cannot serve to argue for substantial change in this regard. The same applies to the noted absences: SOGIE considerations are prevalent neither before nor after the introduction of the goals. Notably, however, one set of NDAC documents, not included in the collocational comparison presented above, seems to paint a different picture: the policy documents, which as explained above, were all produced after 2015. In those texts, both identity and expression collocate with gender (ditto AntConc settings, see footnote 1). The relevant concordance indicates that the association can partly be ascribed to extensive lists or enumerations. CARE’s 2030 Vision (2020, p. 5), for instance, states: “In all contexts, we work with people that are marginalized due to their ethnicity, race, disability, health status, caste, religion, age, class, occupation, history, or sexual orientation, gender identity or expression.”

Such statements, however, are overall of limited frequency, and the strength of collocations such as gender identity in the corpus can mostly be ascribed, once again, to a single document, namely Redd Barna’s Sexual Orientation and Gender Identity and Expression (SOGIE) policy (Save the Children, 2019), which includes a glossary of terms that considerably inflates the relative visibility of particular gender-related patterns in the subcorpus. At this juncture, it is interesting to note that the subcorpus also includes a gender equality policy by the same organisation, which finds very little use for the terminology employed in the above mentioned SOGIE policy. This absence is explicitly explained in the following two sentences:

“When we talk about sex categories, we often talk about women and men, boys and girls, for simplicity’s sake and because many gender norms are built around these traditional categories. It is important to remember that there are many diverse gender identities that do not fit into these categories” (Save the Children, 2017).

The statement serves simultaneously to signal awareness of gender identity, and to dismiss it as a concern of immediate relevance. The document then continues to list the organisation’s guiding principles, in a formulation that also occurs in texts produced before 2015, which once again seems to indicate that the SDGs did not fundamentally affect development discourse circulating in Norwegian aid organizations.

However, absence of evidence does not constitute evidence of absence, and assuming minimal change introduces further questions. Policies and strategy documents operate on a relatively abstract level, and it is therefore possible for organisations to retain a stable discursive core, while producing, in response to external demands, added statement on matters such as SOGIE or the SDGs that do not have a profound effect on the organization’s overall communicative habits. The situation is different, however, for the programmes included in the NDAC: they refer to concrete situations and planned activities, and in this sense the NGOs have to produce a single representation of their aims. An organisation may well have two separate and barely integrated policies on gender identity and gender equality, but programme descriptions must establish priorities. In this respect, there still remains the question of the very high frequency of SDG in the programme texts. Texts are produced to be structurally coherent, and we cannot assume that programme documents can, at the same time, topologically accommodate the insertion of the SDGs—regardless of whether they are restricted to a framing function—without altering, on some level, the general conceptual landscape. In other words, the position claimed by the SDGs in the discursive network of development aid programmes must have already existed if their insertion has only a limited impact on the overall constitution of the discourse. While the Tables above do not immediately clarify the nature of the relevant adaptation, the LancsBox’ GraphColl tool (span 5 < > 5, MI3 measure, frequency threshold of 15, lemmatised and filtered on noun class)^{Footnote 2} highlights an interesting temporal shift in the conceptual network around the noun phrase gender equality, which is a very strong collocation across all NDAC subcorpora (Fig. 3).

Collocates of gender equality in pre-2015 and post-2015 programmes.

Full size image

A number of things are notable in the contrasted graphs, for instance the close association, after the introduction of the SDGs, of violence not just with gender, as discussed above, but also with gender equality. One can note the post-2015 association with specific SDG targets, and the link with other general concerns of Agenda 2030, such as poverty and education. Of immediate interest to the present discussion, however, is the disappearance of three collocates after 2015: human, rights, and development. Based on the graphs, one could argue that the SDG framework has not only subsumed most potential articulations of development under its umbrella but has also replaced immediate reference to another set of globally promoted norms, namely the human rights agenda. This observation should be qualified carefully: the collocation ‘human rights’ appears very often in documents produced after 2015, and we do not argue, therefore, that this framework is at all times replaced by references to the SDGs. Rather, it seems to be the case that the SDGs have come to fulfil a particular discursive function in relation to gender equality that was previously reserved for human rights (rather than for the Millenium Development Goals (MDGs), for instance, which are barely present in the NDAC). This particular conceptual alignment merits further research outside of this article’s scope, as it is no longer strictly related to the position of gender in Norwegian development aid documentation. In the concluding section, however, we will shortly discuss its potential implications.

It is undeniable that the Sustainable Development Goals have gained exceptional global visibility and are frequently appealed to in a variety of sectors, such as education, politics, and corporate communication. The extent and nature of the discursive impact attributed to the Agenda, however, is seldom clearly articulated, and in this study we aimed to examine whether the new framework exerts a salient transformative impact on a particular concept, namely gender, within the domain of development aid documents functioning in the Norwegian context. We employed ‘genre’ distinctions as a comparative benchmark to establish the expected extent of change within the varied communicative settings that NGOs routinely navigate. Our study indicates that, in relation to the concept of gender, no extensive transformation can be noted. The concept’s discursive embedding remains relatively stable across the temporal divide marked by the introduction of the SDGs.

Perhaps this result is not surprising. The Agenda 2030 and the SDGs came about as the result of a complex, UN-led process of consultations, and development aid NGOs were among the many stakeholders involved in civil society consultations. It is not illogical that the influence of NGOs on the formulation of the SDGs enables discursive stability in the development aid documents they produce. A similar point can be made in relation to our final observation, namely the function of the SDGs to occupy a particular discursive position previously held by human rights. Indeed, human rights figure prominently in the preamble to the SDGs, to the extent that one could argue that “Agenda 2030 is conceptually based in human rights” (Arts, 2017, p. 59). As noted in the previous section, however, we do not encounter a full conceptual substitution: the SDGs only fulfil particular communicative functions previously reserved for mentions of human rights. Dembour (2010) lists four complementary interpretations of human rights, which could be helpful in interpreting the observed shift: human rights can be seen as natural entitlements, as agreed upon principles, as contested aspirations, or as meaningless signifiers. The SDGs, which are explicitly framed as goals, can conceptually occupy the aspirational viewpoint, but consequently have to navigate pressures earlier encountered in the context of human rights discourse, particularly as regards the tension between utopian ideals and their practical implementation (Donnelly, 2003).

The focus on targets and indicators in relation to the SDGs can be considered a partial response to this tension, as it ensures that the SDGs include directions for the implementation and assessment of particular programmes. The concurrent procedures of categorization and specification that are part and parcel of Agenda 2030, however, also involve the logical consequence of prioritisation: what cannot be represented as data is not fully taken into account. While the SDGs pledge to leave no one behind, the relation between quantitative monitoring and the increased invisibility of particular types of vulnerability is a commonly raised concern (Davis, 2017). The issue takes on special significance, however, in the context of gender conceptualization. Calculation favours the binary, and as explicitly mentioned in the corpus, SOGIE perspectives are presented as secondary to women’s empowerment concerns, “for simplicity’s sake” (Save the Children, 2017). It is important to stress, however, that this issue is not framed in terms of an either/or dichotomy affecting the entirety of development aid discourse, but rather in terms of particular situated articulations. The concept of gender shows ample flexibility, and restrictions on or transformations of its scope arise in response to particular communicative demands that may arise in response to novel normative agendas, but just as well in recognition of existing institutional expectations or practical considerations. It thus remains an open question whether the SDGs actively inhibit the momentum of discursive developments that operate outside of its focus on measurable targets and indicators, including those that focus on SOGIE as part of gender and development programming.

In this article, we have given due prominence to discursive continuity, in recognition of the fact that, all too often, corpus approaches to discourse accentuate “difference and change,” while underplaying the importance of “similarity and statis” (Taylor, 2018, p. 19). Our study thus also contributes to ongoing debates in CADS about how best to capture and represent the impact of language as social action (Mautner, 2022). In our view, the observation of difference is only meaningful in the context of a study design that will potentially indicate its absence. Collocation analysis thus provided our main point of reference, not only because collocations facilitate unproblematic interaction with concordance lines related to particular concepts, but also because alternatives, such as keyness analysis, can disproportionately accentuate differences between collections of texts that are, in general, quite alike. If continuity, similarity, and statis constitute worthwhile observations, however, they must also be explained, and an important role, in this regard, is reserved for the study of intertextuality as feature of central importance in the analysis of discourse (Rear and Jones, 2013; Farrelly, 2020). Gender, for instance, is embedded in gender equality, which is embedded in gender equality and women’s empowerment, a clause that circulates across a variety of environments that are, in turn, deeply embedded in a myriad of interacting linguistic patterns and practical preoccupations.

While addressing the relationship between discourse and social organization, Stubbs (2010, p. 33) has argued that “[s]ocial institutions and speech acts are the same thing looked at from different points of view.” This parallelism underscores the importance of studying global discursive interventions such as the SDGs, and the importance of linguistic aspects of governance frameworks more generally. It is less immediately clear, however, how one establishes the boundaries of particular speech acts, and therefore, of any institution, given that language patterns are always studied in medias res. Continued methodological reflection is thus needed regarding the methods we employ to capture discursive effects. We hope that our study, beyond its immediate focus on a corpus of Norwegian development aid documentation, also contributes to broader considerations of the interactions between corpus construction, collocation networks, and the assessment of conceptual change.