Discourse and power in the institutionalisation of corporate social responsibility (CSR): A comparative perspective

Abstract This article examines how companies use discourse in corporate social responsibility (CSR) self-reporting to construct their engagement in social issues. Discourse is examined through the lens of “interpretative repertoires” used in the reporting, which appear as clusters of terms that are used to construct versions of reality and thus are constitutive of CSR. Using an innovative perspective that combines discursive institutionalism with Foucault’s notions of knowledge, power and discourse, the article examines how interpretative repertoires are used in self-reporting and how this has social consequences. The contribution of this article to CSR research is both theoretical and empirical. Beginning with the assumption that context is an important aspect of how discourse is constructed, the empirical material includes company self-reporting from two emerging economy contexts (South Africa and Mauritius) and two advanced welfare states (Sweden and the UK). The analysis reveals the ways in which interpretative repertoires reflect how the versions of CSR are anchored in institutional contexts. How these repertoires are constructed also reveals how companies exercise power by constructing the conditions, and setting the boundaries, for company engagement in social issues.

ABOUT THE AUTHOR Brita Backlund Rambaree (PhD) is a lecturer in social work and sociology at the University of Gävle and currently the Programme Coordinator for the university's Bachelor programme in International Social Work. She holds a PhD in Sociology from Stockholm University with a thesis on the topic of CSR. Her main research interests concern how perspectives from social work, social policy and sociology can be applied to CSR.

PUBLIC INTEREST STATEMENT
This article examines how companies use discourse in corporate social responsibility (CSR) selfreporting to construct their engagement in social issues. Discourse is examined by looking at "interpretative repertoires", which are recurring sets of explanations that appear in the selfreporting. By focusing on interpretative repertoires the article contributes to the debate on CSR with an analysis of how power and knowledge intersect as companies through self-reporting construct the understandings of CSR that come to prevail. The contribution of this article also lies in its illustration of the relevance of the national context for how CSR is discursively constructed. As the empirical findings show, how companies construct and uphold the meaning of CSR is, at least in part, shaped by the national context.

Introduction
Corporate social responsibility (CSR) is a term widely used in academic research and public discourse when referring to the social aspects of doing business. In this sense, it is an umbrella concept that covers various expressions, such as stakeholder engagement, sustainability and corporate citizenship (Strand et al., 2014). How companies engage in social aspects is communicated in self-reporting on CSR, and such reporting is found in specific CSR, sustainability or corporate responsibility reports, as well as in annual reports, on company websites and through other media. CSR self-reporting has been growing, and the production of substantial reports dedicated to CSR has become ever more common as the concept has increasingly been adopted by the corporate world (Owen & O'Dwyer, 2008;Waddock, 2018;Wang et al., 2018).
The increasing involvement in CSR and its reporting calls for research that engages with discourses on CSR. As companies report on CSR, they both draw on existing discourses on CSR and contribute to producing and reproducing discourses that define CSR and its boundaries. Hence, how companies use discourse in self-reporting to construct their engagement in social issues bears relevance for the versions of CSR that become institutionalised and their meanings widely accepted.
The aim of the article is to examine how CSR is discursively constructed in self-reporting, and what the similarities and differences are in the discursive constructions among large companies in four different national settings across two distinct global regions: Western Europe and Southern Africa. Using the concept of discursive institutionalism, which assumes that institutions are "social constructions constituted through discourse" (Phillips et al., 2004, p. 638), I argue that discourse is a key component in understanding how CSR as an institution is produced and reproduced within companies and the broader society (Phillips et al., 2004;Phillips & Malhotra, 2008;Schmidt, 2008). Michel Foucault's perspective on power in discourse (Foucault, 1980(Foucault, , 1982, integrated within the framework of discursive institutionalism, provides the conceptual leverage to understand the processes by which specific understandings of CSR come to be accepted and taken for granted. The critical literature on CSR has shown that CSR promotes understandings of companies' role in society that are corporate-centred, restrict issues to those that suit company perspectives, and do little to address the negative aspects of business activities (see for instance, Banerjee, 2008;Fleming & Jones, 2013;Nyberg et al., 2013;Schneider, 2020). Under the banner of CSR, companies communicate the idea that their actions imply an alteration of practices in the global capitalist system whereas in fact CSR can be seen to uphold unchanged and unsustainable practices (Banerjee, 2008;Hanlon, 2008;Valentin & Murillo, 2011;Waddock, 2018). As Fleming and Jones (2013) point out, CSR is a discursive system promoting the notion that it entails a solution to the world's problems. Precisely how this discursive system is upheld through CSR reporting is the focus of this article and this is studied by drawing on Foucault's work on power and knowledge in discourses. Acknowledging that discourse is contextually constructed, not just in relation to the capitalist system, but also in (among others) national contexts, the article analyses the reporting in relation to differences across countries. In this study four countries with large differences in sociopolitical terms were purposively selected to provide the setting in which to study the contextual construction of CSR. As the construction of CSR takes place in a specific social context it is useful to conduct an analysis that explores differences and similarities across contexts.
More specifically, this article contributes to the critical literature on CSR and research on discourse in CSR by analysing the use of "interpretative repertoires"; that is, "relatively internally consistent, bounded language units" (Wetherell & Potter, 1988, p. 171) that are used in companies' self-reporting and are constitutive of CSR's meanings. This approach allows for a nuanced and empirically grounded analysis of how the meaning around CSR is constructed in the studied self-reporting in four country contexts, and how these discourses legitimate company power. The article then also contributes to institutional discourse theory by showing how interpretative repertoires are used in context-oriented ways to construct the meaning of CSR. Thus, the article would be relevant to scholars and students interested in critical CSR studies, discourse in CSR or discursive institutionalism. It could also be relevant for practitioners involved in monitoring or developing CSR reporting as it provides insights on building credible reporting.
The article is divided into six sections. After this introductory section, the theoretical frameworks are outlined, followed by a discussion of contexts. In part three the methods are outlined. Part four presents the analysis of interpretative repertoires and how they are anchored in the four contexts. In part five the findings are discussed and in the final section conclusions and implications for future research are presented.

CSR through the lens of discursive institutionalism
Approaching CSR through the lens of discursive institutionalism highlights discourse as an essential aspect of the social construction of institutions (Phillips et al., 2004;Phillips & Malhotra, 2008;Schmidt, 2008). With a discourse analytical approach, companies reporting on CSR are seen as using language to "create representations of reality that are never mere reflections of a preexisting reality but contribute to constructing reality" (Jørgensen & Phillips, 2010, pp. 8-9). As such, the discourses contained in the CSR self-reporting contribute to producing what a company is, what CSR is, and what type of social issues companies can be expected to engage in. From the perspective of discursive institutionalism, discourses influence the understandings and definitions of reality that come to prevail in a field (Phillips et al., 2004, p. 638).

On power in analysing company discourse
Seen through Foucault's lens of power in discourse, these discourses in companies' self-reporting reveal how companies set boundaries for what is and is not possible; what can and cannot be expected. As Foucault (1980Foucault ( , 1982 has discussed at length in his work on knowledge, power and discourse, power is not just repressive but also productive, and in the latter sense entails the production of knowledge and discourse. Accepted knowledge, "truths", are created in discourses and give impressions of true and false representations of reality (Foucault, 1980). "Truth is linked in a circular relation with systems of power which produce and sustain it, and to effects of power which it induces and which extend it" (Foucault, 1980, p. 133). In other words, companies' exercise of power, in a productive sense, involves the production of knowledge in self-reporting. The knowledge produced in self-reporting contributes to constructing broader discourses on CSR and the role of companies. These discourses are then constructive of how CSR is perceived and how further knowledge is produced. The discourses that are constructed shape what is considered to be meaningful, true and false, and set the conditions for what is possible (Foucault, 1980(Foucault, , 1982Jørgensen & Phillips, 2010). Alternative constructions of social issues and the role of business in society that may be less beneficial to companies become marginalised, as they are kept outside the discourse on CSR. A schematic representation of the interrelations between power, discourse and production of knowledge in CSR self-reporting is presented in Figure 1. As the bidirectional arrows indicate, the interrelations are continuous and complex between power, knowledge and discourse.
Foucault offers a theoretical point of departure for considering the role of power in CSR discourses. However, to be able to operationalise the discourses contained in the self-reporting, this article uses an innovative approach that combines Foucault's theoretical approach with a more micro-level analysis of interpretative repertoires.
The idea of interpretative repertoires has been developed by Potter and Wetherell (Potter & Wetherell, 1987;Wetherell, 1998Wetherell, , 2006Wetherell & Potter, 1992) as a central concept in the approach to discourse analysis known as discursive psychology. Albeit similar to Foucault's notion of discourse, interpretative repertoires are less abstract and allow us to consider how language is used flexibly to perform social actions (Jørgensen & Phillips, 2010;Wetherell, 2006). While a discourse can be seen as "a particular way of talking about and understanding the world (or an aspect of the world)" (Jørgensen & Phillips, 2010, p. 1), interpretative repertoires are recurring sets of explanations, "clusters of terms organised around a central metaphor", and are used in talk and text to construct reality (Wiggins & Potter, 2008, p. 74). More specifically, they can be described as "recognizable routines of connected arguments, explanations, evaluations and descriptions which often depend on familiar anecdotes, illustrations, tropes or clichés" (Wetherell, 2006, p. 154). For instance, when Wetherell and Potter (1992) studied discourses on racism in New Zeeland, they analysed a "race repertoire" as well as a "culture repertoire" as different ways of talking about differences and examined how these interpretative repertoires are used as resources that people draw on to construct reality. Thus, the idea of interpretative repertoires emphasises that discourses are actively used as flexible resources in order to accomplish social actions within specific contexts (Jørgensen & Phillips, 2010). In this way, the focus on interpretative repertoires implies an extension on the perspective of Foucault. Foucault's work makes visible the dimension of power in the production of knowledge and discourse on CSR and focuses on the effects of discourse on people, but an analysis of interpretative repertoires enables a focus on both how discourses affect our understandings of the world and how discourse is affected by how we actively put it to use. An analysis of interpretative repertoires then connects the active use of language to the social consequences of discourses. For instance, interpretative repertoires can be used in text and talk to construct versions of reality that uphold inequalities (Jørgensen & Phillips, 2010;Wetherell & Potter, 1992).
In a CSR context, identifying interpretative repertoires involves locating themes and meaningful stories that companies draw on to construct meaning around CSR. In other words, interpretative repertoires are taken as "discourse use in practice" (Wetherell & Potter, 1992, p. 92). As the concept of CSR has spread globally the interpretative repertoires in company self-reporting can be expected to be similar across the globe, but at the same time, differences in approaches to CSR across contexts suggest that these repertoires may also differ (see for instance, Cantó-Milà & Lozano, 2009;Sobczak & Coelho Martins, 2010;Vertigans et al., 2016). The construction of company CSR would take place in a specific context and it is therefore useful to explore differences and similarities across contexts. Company constructions of CSR would however also connect to earlier discourses and the use of interpretative repertoires can be understood as a process of actions and reactions resulting from social interactions. Thus, while this study is limited to discourse at a particular point in time, the historical context would also be important and can in future research contribute to understandings of how the interpretative repertoires have emerged.

Anchoring CSR in an institutional context
Given that CSR is a global concept with shared ideas and practices promoted by organisations such as the UN and the ICC (Gjølberg, 2009), CSR discourse can be perceived as global and relatively homogenous. At the same time, however, CSR self-reporting, like other forms of talk and text, is oriented towards action in specific contexts (Jørgensen & Phillips, 2010;Wiggins & Potter, 2008) and CSR would, importantly, also differ across contexts. Countries differ in socio-political terms, whereby historical, political and economic developments are likely to interact with the creation and development of CSR discourse. Particularly large differences in socio-political terms exist between emerging economies and advanced welfare states.
The empirical material analysed in this article includes 40 leading companies from two West European advanced welfare states (Sweden and the UK) and two Southern African emerging economies (South Africa and Mauritius) where CSR is prevalent. Although the advanced welfare states and emerging economies are two distinct contexts in many ways, contextual differences in CSR also exist within them when Sweden is compared with the UK, and Mauritius with South Africa. Sweden provides a country context where extensive welfare provision has been linked to an "implicit" approach to CSR, where responsibility is built-in in broader policy arrangements (Matten & Moon, 2008). This differs from the "explicit" articulation of CSR that has longer prevailed in the UK liberal economic environment (Matten & Moon, 2008). Similarly, Mauritius and South Africa are also two separate contexts, with CSR in Mauritius having taken on a much more state-directed and non-voluntary approach. Thus, the countries in this study differ as advanced welfare states versus emerging economies, but also as four separate socio-political environments with differences in how CSR is used. In this study, these contextual differences are used in exploring similarities and differences in how CSR is discursively constructed in company self-reporting. In what ways are the discourses universal and can be seen to reflect a similar advantage of corporate power across the different contexts, and in what ways do they differ and represent different ways of constructing company engagement in social issues?

Four national contexts
Two advanced welfare states (Sweden and the UK) and two emerging economies (South Africa and Mauritius) provide the country contexts in this study of CSR's discursive constructions. As advanced welfare states in Western Europe and emerging economies in the Global South, these two contexts differ widely in the capacity of formal institutions such as social insurance, social services and health care services to ensure welfare and address social issues outside CSR. This can be expected to matter for how companies engage in CSR and construct its meaning. As Visser (2008: 474) has pointed out, emerging economy contexts "present a distinctive set of CSR agenda challenges". Given the greater development needs, CSR would tend to have a greater focus on issues related to social development within the country. Compared to other non-Western and non-OECD country contexts, South Africa and Mauritius are interesting cases for CSR research given that both are emerging economies with corporate environments in which CSR is commonplace; in addition, both have relatively established welfare systems that address social issues outside the framework of CSR. Compared with Sweden and the UK they have a context of greater development needs, which can be expected to be reflected in CSR discourses.
These two global contexts also differ along the dimensions of voluntary and mandatory frameworks for CSR; that is, the extent of government steering of CSR activities. Sweden and the UK represent the voluntary framework in CSR, in which companies largely choose how to engage in CSR. While Sweden recently took a step towards increased government involvement through the introduction of mandatory CSR reporting for all large companies as of December 2016, in both countries CSR entails activities that are largely defined by the companies themselves. South Africa and Mauritius, on the other hand, represent a context where government initiatives have shaped the CSR content, and are country cases that depart from the widespread notion that CSR concerns responsibilities that companies voluntarily choose to engage in.
At the same time, the four countries in this study also differ within their regional contexts. As such, the UK is considered a global leader in CSR and is a context where companies tend to communicate extensively on CSR activities, which has been attributed to the liberal orientation of the welfare state with restricted public provision of welfare (Gjølberg, 2009;Kinderman, 2012;Matten & Moon, 2008). Sweden, on the other hand, represents a country context where it has taken longer for the concept of CSR to gain ground, and where CSR covers a more restricted set of issues, which has been attributed to an extensive welfare state (De Geer et al., 2009;Roome, 2005).
South Africa and Mauritius have also taken different trajectories in CSR, with Mauritius taking a more interventionist approach than South Africa and introducing mandatory CSR in 2009. In both countries, government initiatives have played a role in the development of CSR. In South Africa, the government initiative Broad-Based Black Economic Empowerment (B-BBEE), with focus on aspects such as human resource development, employment equity, supply chains and preferential procurement, has shaped CSR as it indirectly compels companies to engage in responsible practices (Ramlall, 2012). However, government initiatives have been much more central in the development of CSR in Mauritius, with mandatory regulations directing it. In Mauritius, every profitable company has to dedicate two per cent of its annual profits to CSR, and for the period covered in this analysis (years 2010/2011) this contribution had to be directed at social and environmental development within the country (National Empowerment Foundation (NEF), 2011). 1 While Mauritius can be described as having a greater focus on universal welfare provision than South Africa (see for instance, Sandbrook et al., 2007;Seekings, 2005;Willmore, 2006), in both countries CSR has come to supplement state welfare, given the context of gaps in welfare provision (see Fig, 2007;Hamann, 2009; National Empowerment Foundation (NEF), 2011). Thus, while representing four different national contexts, these countries also represent two distinct regional contexts as established and emerging economies with clear differences in CSR agendas. With the selection of countries in this study, the contextual differences in background and approaches to CSR are useful for exploring similarities and differences in how CSR is discursively constructed. However, these are cases that allow us not only to provide points of comparison, but also to broaden the conceptual terrain.

Data collection
Large companies tend to be "agenda setters in CSR", and have the resources to put into CSR (Chapple & Moon, 2005, p. 424). As "agenda setters", the large companies can influence what is included in the CSR agenda and thereby direct attention to specific constructions of CSR. Hence, ten of the largest incorporated companies in each country were selected. The selected companies all have operations in more than one country. The sampling was based on listings from the national stock exchanges for the Swedish, UK and South African sample, and on a published ranking of companies for the Mauritian sample. 2 Subsidiaries and companies not incorporated in, or not having headquarters in, the countries studied were excluded from the sample, as were companies that did not have self-reporting available on their website.
The companies' CSR reports for 2010 or 2010/2011 were analysed. If a specific CSR report was not available, sections of the annual reports that addressed CSR were used instead. For the UK sample, eight of the selected ten companies had a separate CSR report. Among the Swedish and South African companies, seven of the ten had published a separate CSR report. The Mauritian companies only reported on CSR in their annual reports. Table 1 provides details on company sectors for each country. Some of the companies operate in sectors that are more controversial than others, such as those dealing in tobacco and extractive materials. In the analysis, particular aspects of self-reporting by these companies are pointed out.

Methodological issues and strategies
The benefits of studying self-reporting as a reflection of company behaviour may be questionable since the reporting may differ greatly from how CSR is practiced, considering how CSR is often a means of managing the company image and offsetting criticism regarding social impacts (see for instance, Waddock, 2018;Munshi & Kurian, 2007;Blowfield & Frynas, 2005;Utting, 2007;Waddock, 2007;or Zorn & Collins, 2007). Departing from the notion of self-reporting as recording social reality, this study uses CSR reports as sites for the formulation and articulation of discourses in different contexts. The CSR reports are interesting because of (and not despite) their connections to the corporate environments in which they were created. The reports reflect these environments, but also play a role in constructing and changing them. One could therefore see these reports as "media through which power is expressed" (May, 2008, p. 183).

Analytical strategies
The analysis of CSR self-reporting focused on identifying interpretative repertoires, as expressions of "discourse use in practice" (Wetherell & Potter, 1992, p. 92). This involved a lengthy process of going back and forth between report-reading, coding and the themes that were developed. A strategy of looking for similarities as well as differences in the reporting across countries and over-all in the sample was used, and codes were allocated to the text with the help of the qualitative data analysis software ATLAS-ti (version 6.1). The process of assigning codes involved merging codes with similar meaning, thus attaining broader codes that were grouped into draft interpretative repertoires. The reporting was then read again to check for coherence in these interpretative repertoires. Table 2, below, provides examples of how codes were assigned to quotes from the texts, and how these were then grouped under interpretative repertoires. The analysis focused on social aspects, which meant that information about environmental responsibilities (involving issues such as CO 2 emissions or water usage) was left out. Although environmental and social responsibilities overlap as aspects of sustainability, it can be useful to consider social aspects separately since companies may use other types of interpretative repertoires when reporting on environmental aspects. In all, four recurring interpretative repertoires were identified in the company self-reporting: asserting the business motive, making a difference, the win-win situation and being part of the solution. These interpretative repertoires reflect CSR constructions as communication targeting a broad audience-including, for instance, investors, suppliers, employees, media and community groups, with potentially very different interests in the company's activities. These four interpretative repertoires are the resources that companies draw upon in order to position themselves, within the universe of CSR discourse, as capable of enhancing financial profitability and of contributing to a better world.  (3) retail (1) IT (1) telecom (1) energy (2) extractive materials (2) financial services (2) pharmaceuticals (2) telecom (1) tobacco (1) extractive materials (3) financial services (3) telecom (1) media (1) energy (1) consumer staples (1) multiple (6) travel & tourism (3) telecom (1) Note: The category multiple refers to companies with business operations in multiple sectors, such as financial services, tourism, logistics, commerce, agricultural and industrial production.

Findings: interpretative repertoires
The four interpretative repertoires are interrelated (see Figure 2), and were used as resources in the discursive construction of CSR. Taken together, the interpretative repertoires produce a discourse that sets the conditions for company engagement in social issues; in this way, they entail the exercise of power by constructing the meaning of "truth" (Foucault, 1980(Foucault, , 1982. The discourse that is produced shapes what can be seen as meaningful, relevant and possible company involvement. As such, discourse on CSR contributes to producing what CSR is and what it can reasonably be expected to cover. In other words, the exercise of power, in a productive sense, by companies, involves the production of knowledge in self-reporting. As Figure 2 shows, the discourse that was promoted reconciles two rather different logics: making a difference in society and company financial growth. While the former suggests a rationale for CSR beyond self-interest and with the greater good of society in mind, the latter suggests a rationale based on company financial benefits. Both were interpretative repertoires recurring in the selfreporting as companies explicitly affirmed the importance of the business motive, but at the same time emphasised making a difference in society beyond profit-making. By asserting both aspects and joining them through the interpretative repertoire of win-win situations, the companies could stress (as a fourth interpretative repertoire) that they are part of the solution to social problems. This merging of the two logics of earning a profit and making a difference in society meant that CSR was defined as those issues that benefit the company as well as the broader society. The merging of logics, understood within a Foucauldian perspective, reveals how "truth" about CSR can be framed as a win-win situation, which effectively deflects from any tensions that may exist between the two logics.
All four interpretative repertoires were prominent in the reporting of the Swedish, UK and South African companies, which suggests that they are part of contemporary understandings of CSR and are applied across contexts. Assertions of the business motive and win-win situations were not present in the Mauritian CSR reporting as interpretative repertoires, since the mandated CSR implied less need to justify involvement in CSR activities in terms of contributions to profitmaking. Instead, the reporting drew only on the repertoires of making a difference and being part of the solution, in order to promote the company and its CSR efforts. In the sections that follow, the use of each of the four interpretative repertoires is analysed.

Asserting the business motive
The Swedish, UK and South African companies linked their CSR activities to their core business, and also frequently made reference to the benefits to business from engaging in CSR. Given that a business motive bears associations to profit-making, self-interest and greed, it appears somewhat paradoxical that the avowal of a business rationale is used in the presentation of company social responsibility. However, as a "vocabulary of motives"-that is, how language is used to justify motives (Mills, 1940)-the business case argument is often accepted, and even valued, when associated with sustainability in the discourses on corporate responsibility. 3 By stating the business case for CSR involvement, the companies can show that they are genuinely interested in, and have a sustained commitment to, the issues they report on.
For instance, in the reporting of the UK-based energy company BG Group (2011) and mining company Rio Tinto (2011), the link to business interests was clear and explicit as the companies mentioned that involvement in local communities and support from them are important for their future business prospects. Similarly, the UK telecom company Vodafone Group Plc (2011, p. 3) stated that meeting challenges of development in emerging markets provides opportunities for business growth: "The future of our business is inextricably linked with global sustainability challenges. We are targeting the areas where our mobile solutions can help to meet these challenges at the same time as developing new opportunities for business growth".
The interpretative repertoire of asserting the business motive was also evident in the Swedish companies' reporting. For instance, the reporting of the retail company H&M (2011, p. 100) mentioned that the company contributes to economic growth as well as improved social and environmental conditions, and that "Investing in the communities around us extends these improvements more widely and in turn, benefits our business".
The business rationale was also found in South African reporting, where CSR is partly driven by a state-developed national framework for inclusion (the B-BBEE initiative) that indirectly compels companies to engage in responsible practices. For instance, the supermarket chain Shoprite Holdings Ltd (2010) stated that support to local communities earns the company support from customers, suppliers and potential employees. Similarly, mining companies AngloGold Ashanti (2011), Gold Fields Limited (2010), and Implats (2010) all made reference to the beneficial outcomes for business from committing to engagement in the communities where their mines are located. In the case of Standard Bank, the reporting included the following statement about the business imperative of engaging with people in the informal economy: 'The informal economy accounts for millions of economically active people in Africa who fall outside the mainstream financial sector. Overlooking this marginalised sector of society would not only affect our sustained profitability in emerging markets, but also the broader socioeconomic development potential of the countries in which we operate'. (Standard Bank Group Limited, 2011, p. 25) Presented in this way CSR is based on the logic of sustaining company financial growth and the business motive is then the rationale for CSR involvement.
Such constructions of CSR, that link CSR to company business operations and point to the financial benefits of CSR involvement, were not present in the Mauritian CSR reporting. The mandatory nature of CSR in Mauritius imply less need to justify involvement in CSR activities in terms of contributions to profit-making. Being legally obligated to be involved with CSR means that the companies do not need to justify why they are allocating resources to these types of activities. The context has then influenced the CSR approach within the texts.
In other contexts, asserting the business motive can be a justification for CSR involvement. In the South African reporting, the business motive was for instance, evident in the use of the term "corporate social investment" (CSI). The use of CSI instead of CSR signifies that "social and environmental interventions by firms are seen as investments with the purpose of some kind of return in the form of reputation, enhanced markets or brand recognition" (Fig, 2007, p. 8).
The interpretative repertoire of asserting the business motive was also evident in how the companies linked risk management to the meanings of responsible operations. Issues such as health and safety, sourcing, supply chains and product quality were discussed by the Swedish, UK and South African companies in terms of potential risk to the company. 4 For instance, the Swedish industrial company Sandvik stated: 'To raise the level of understanding in the organization regarding the risks relating to human rights and labor issues when Sandvik undertakes operations in high-risk countries, analyses have been conducted of a large number of countries in Asia, Africa and South America'. (Sandvik, 2011, p. 101) The construction of responsible production as risk management is a construction that is in line with the emphasis on the business rationale. The South African chemical products producer Sasol (2010, p. 16) specified: "Through effective risk management practices that prevent incidents, we save on potential cleanup costs, insurance premiums and legal liabilities, not to mention the intangible costs associated with an impaired reputation".

Making a difference
In addition to the interpretative repertoire of asserting the business motive, an interpretative repertoire of making a difference was also visible in the CSR reporting. This repertoire appeared across all four country cases in how the companies constructed community involvement as activities that have a substantial and lasting impact on society, and often as activities that surpass charitable giving. For example, Swedish IT company Ericsson (2011, p. 4) stated: "Our work is not about philanthropy. For lasting change, there has to be a sustainable business case". The other reports did not distance themselves from philanthropy in the same way, but many conveyed the message that their community involvement is concerned with contributing to local communities by addressing existing problems and promoting development. Pharmaceutical company GlaxoSmithKline (2011, p. 248) (UK) stated: "We aim to make a real difference to communities by working with our partners to find innovative solutions to healthcare challenges". Standard Bank Group Limited (2011, p. 40) (South Africa) stated: "Beyond donating money to charities, we are in a position to provide skilled support to organisations that is sustainable and offers long-lasting benefit". The CSR involvement was presented as activities and endeavours that make a difference and have an impact on society.
The same interpretative repertoire of making a difference was present across all four countries but among the Mauritian and the South African companies, the focus was more specific to the context of the country of incorporation. The Mauritian company Sun Resorts Limited (2011, p. 25), for instance, stated: "Our commitment towards corporate social responsibility programmes at both local and national levels, in line with Company objectives, has proved to be a very powerful tool in bringing improvements to local communities and disadvantaged sectors of the population".
Given that the Mauritian companies largely reported on activities that lay outside their business areas, their CSR activities could fall within the category of charitable endeavours. Within the CSR reporting, however, they were cast more in terms of commitment to social development, and as such, as more long-term contributions.

Credible involvement
Across the four countries, company reporting also often made reference to how the companies have used assessments, surveys, evaluations, plans, consultations, focus group studies or similar as methods to develop the CSR involvement with the community. The reporting drew on a wider discourse about the relevance of research-based approaches and scientifically founded knowledge. A reporting that presents knowledge gained through such systematic techniques of knowledge production has the potential to be more influential, as it strengthens and lends credibility to the company's CSR involvement. As Smart, quoting Foucault, writes: "Mechanisms of power have been accompanied by 'the production of effective instruments for the formation and accumulation of knowledge-methods of observation, techniques of registration, procedures for investigation and research, apparatuses of control'" (Smart, 2002, p. 80). Thus, referring to the use of research further adds to the power of large companies to construct the conditions for their involvement in social issues, especially since large companies generally would have the resources to commission studies. For instance, among the South African companies, Standard Bank Group Limited (2011) mentioned using a research-based approach to understanding needs in the community, and Implats (2010) pointed out that they have a dedicated team for the implementation of socioeconomic projects and that they use independent impact assessments. Similarly, the Swedish company SKF Group (2011) described how they use "community care plans" to assess to the needs of local communities.
Such descriptions underscore the extent to which CSR discourses can frame the image of company activities as involvement that goes beyond ad-hoc charitable activities and has an impact on the social needs of the community. The interpretative repertoire of making a difference, supported by assertions of credible involvement give weight to company CSR involvement.

Win-win situations (the merging of two logics)
The two logics, financial growth and making a difference were merged into a third interpretative repertoire: win-win situations. Particularly among the Swedish and UK companies, CSR was commonly constructed as activities that have the dual purpose of benefitting both business and the wider community. A discourse of CSR as win-win situations was upheld as the companies used an interpretative repertoire that merged the two logics of earning a profit and contributing to society. As an accepted notion in contemporary CSR discourse, the win-win logic lends legitimacy to company engagement in CSR, as it indicates a substantial engagement in it. The interpretative repertoire of merged logics is one that creates a "truth" about profit-making and societal benefits as company aims that go hand-in-hand. These discursive constructions were evident particularly in the way the companies merged core business with societal contributions in their CSR reporting. For example, the Swedish bank SEB (2011) reported on activities such as improving financial literacy and engaging in promoting entrepreneurship. Similarly, the UK financial-sector companies Lloyds Banking Group (2011) and HSBC (2011) both reported working with financial inclusion and financial education. As the report by Lloyds stated: 'We aim to lead the banking sector in reaching those that are financially excluded and equip them with the confidence and capability to manage their money effectively. We also have a strong commercial interest in helping to create a nation of consumers who are both comfortable and confident in dealing with the financial services sector'. (Lloyds Banking Group, 2011, p. 25) The UK-based pharmaceutical companies GlaxoSmithKline (2011) and Astra Zeneca (2010) provided information on how they engaged in extending access to healthcare for communities around the world and also connected the societal benefits of these initiatives to business motives:. 'Ultimately we believe that responsible business is good for society and good for GSK. It helps us to operate efficiently, to gain the trust of our stakeholders, to create the products that patients and healthcare payers really need and to foster the right conditions for expansion of our business'. (GlaxoSmithKline, 2011, p. 8) For the (then) Swedish vehicle manufacturer Volvo Group (2011), the production side's concern with traffic safety was reflected in the community involvement, as the company reported on cooperation with different stakeholders on safety awareness issues and the provision of funding to projects in this field.
A more general and encompassing overlapping of the companies' core business activities and their community involvement was also evident in some of the reports. For instance, the Swedish IT company Ericsson (2011) and the Swedish telecom company Telia Sonera AB (2011) both described their products and services, which extend access to technology and communication services, as contributions to addressing social and environmental problems. In these two reports it was not simply the case that the company's involvement in the community was in line with its business/profit role; the company's contribution to society and its core business were in fact merged so that the business activities per se were presented as contributing to social development. Here, the interpretative repertoire of merged logics was used to construct the image of a complete win-win situation. The Telia Sonera AB (2011, p. 3) stated that "TeliaSonera services drive growth and competiveness and we contribute to a more open and digitally inclusive society by providing possibilities for more people to communicate." (Telia Sonera AB, 2011, p. 3) While it can be argued that the Swedish-and UK-based companies focus their attention on areas where they are best able to make a difference, it is also the case that their constructions set the agenda for CSR by merging self-interest and responsibility (Sklair & Miller, 2010).
The South African companies also linked community engagement with company business activities, but at the same time reported fairly extensively on issues with less direct connection to the companies' operations. For instance, health issues and support for cultural events were presented as part of how the companies engage in communities, which reflects how CSR in the South African context is also concerned more generally with social issues within the country. In some cases also this broader approach to CSR was presented as a win-win situation. The report of the financial services company Sanlam mentioned that their community involvement includes education, environmental and HIV components and "By investing in these areas, Sanlam RSA not only encourages future customer loyalty and support, but we make a meaningful contribution to the society we serve" (Sanlam, 2011, p. 84). The South African CSR reporting also included Broad-Based Black Economic Empowerment (B-BBEE) or transformation in favour of disadvantaged groups. 5 This is a state-developed national framework for redressing the entrenched inequalities existing in the society and it indirectly compels businesses to be involved (Ramlall, 2012). Thus, as the companies also reported on issues that were not derived from their own win-win logic, the discourse of the win-win logic was less encompassing in the South African reporting than in that of the Swedish and UK companies.
The Mauritian companies did not connect CSR to company core operations, instead presenting it as a separate line of company activities, concerned with charitable involvement in local social problems. Given the mandatory nature of CSR and the fact that the government has already specified certain issues to which it should be directed, there would be less room for companies to use it independently in ways that link it to business activities, and less need to justify involvement in CSR activities in terms of contributions to profit-making. Thus, the interpretative repertoire of win-win logic did not have the same salience in the Mauritian reporting as in the reporting in the other countries in the study.

The company as part of the solution
By constructing CSR as a win-win situation that benefits both business and the wider community, companies can stress-as a fourth interpretative repertoire-that they are part of the solution to social problems. The interpretative repertoire of the company as part of the solution to social problems was visible in the CSR reporting across the four country contexts. As an interpretative repertoire, however, it was used differently in the self-reporting among the companies in the two Western welfare states and the two emerging economy contexts. The UK and Swedish reporting largely portrayed the company as a significant actor promoting economic and social development in other regions of the world, whereas the South African and Mauritian companies' reporting emphasised the companies' importance in addressing local social problems and furthering social development within the country. Many of the UK and Swedish companies have operations that are particularly widely spread across the globe. To gain the trust of consumers and investors, and to maintain successful business operations, it is necessary for these multinationals to show that they are contributing to the socio-economic needs of the lower-income countries where they have operations, and not merely exploiting the resources there. Hence, they emphasised this aspect of their operations in other regions of the world. The South African and Mauritian companies in the study also have operations in more than one country. However, since they are incorporated in contexts of greater socio-economic needs, legitimacy can mainly be gained by showing how they are an important part of addressing socio-economic problems within the country. As Hamann (2009, p. 439) has noted, the orientation towards social development within the country of incorporation is often expected of companies in Africa, given the context of great socioeconomic development needs. Thus, the use of the interpretative repertoire of being part of the solution reflects the broader institutional contextual setting; being part of the solution to social problems would have a different meaning in an emerging economy context than in established welfare states.
For instance, the Swedish company SKF Group (2011) mentioned that the company, having more than a third of its employees in Central/Eastern Europe, Latin America, Asia, Middle East and Africa, brings technology and training to these regions and contributes to development there. In other words, it is a representation of the company as a significant contributor, particularly to those regions of the world where it is most needed. As Edwards and Potter (1992, p. 16) have mentioned, descriptions "are constructed to do communicative, interactional work", and in this case the description presents the company as active in local communities and committed to social progress in and through its global presence. The same interpretative repertoire appeared in the selfreporting of other companies as well. The retail company H&M (2011, p. 100) mentioned: "We contribute to economic growth in the areas that H&M and our suppliers operate in and use our influence to promote better social and environmental conditions." Constructions of the company as part of the solution to social problems were also part of the South African and Mauritian reporting. For instance, the South African financial services company Sanlam (2011, p. 88) gave the assurance that "We will continue to utilise our resources to overcome our country's social and environmental challenges". Apart from the focus on the local context, the discursive construction of the role of the company in this quote was the same as among the UK and Swedish companies. Similarly the Mauritian company CIEL Agro Industry (2010, p. 29) mentioned the following about the foundation it has set up for CSR activities, the "Fondation Nouveau Regard" ["New Perspectives Foundation" (author translation)]: 'Over the past five years, it has been involved in community development projects throughout the island, focusing on children in great distress, including those who grow up in the street and those in difficult family situations. With the support of the UNDP, it has brought together three NGOs in order to create a network and share experience, means and capacity for the care of these children'. (CIEL Agro Industry, 2010, p. 29) The quote exemplifies the focus on social problems within the country, as the company positions itself as part of the solution. The involvement, while charitable, should not be viewed as ad-hoc activities or simply donations, and is presented as entailing important contributions in terms of taking the lead in a collaboration that can address a serious social problem within the country.
Some of the companies studied, namely those operating in more controversial sectors such as extractive materials, energy or tobacco, presented themselves as part of the solution to social problems in more narrow terms. This reflects how discourse is situated not just within country contexts but also within a context of specific business sectors. Instead of conveying the image of the company as an actor in promoting development, the reports of UK-based companies BP PLC (2011) and British American Tobacco (2011), as well as the extractive materials companies (two UK-based and three from South Africa), focused on more specific impacts of company operations and products on local communities, customers, suppliers and other stakeholder groups. Presenting, for instance, mining companies as promoters of social and economic development for local communities clearly requires justifications of just how the companies go about doing this in the localities they are often criticised for exploiting. Similarly, in the aftermath of the months-long oil spill in the Gulf of Mexico, in BP PLC's (2011) case being part of the solution meant emphasising that the company swiftly took responsibility for the clean-up. For these companies, the context of operating in a controversial sector shapes constructions of CSR so that the meanings given to CSR perform the "communicative action" (Edwards & Potter, 1992) of legitimating controversial operations (Breeze, 2012).

Similarities in discursive constructions: discourse accomplishing social actions
The interpretative repertoires reveal the resources drawn on by the companies to discursively construct CSR, and illustrate the process of discursively institutionalising specific perceptions of responsibility. The interpretative repertoires reflect how companies produce knowledge and "truths" about what it entails to engage in social issues, and what it entails to report on this engagement. Through their self-reporting, companies set the boundaries for company involvement with social issues. As Foucault (1982, p. 789) has stated, the exercise of power in the productive sense "consists in guiding the possibility of conduct and putting in order the possible outcome".
The four interpretative repertoires that recur in the texts largely overlap, and can be seen as discursive constructions that centre on a broader assertion about companies and profit-making activities as indispensable for society and social development. Although the reported forms of involvement spoke of both more and less committed types of engagement, CSR was generally constructed as activities with a substantial impact. None of these cases exhibited a discourse of philanthropy; instead, the companies portrayed themselves as providers of solutions to social and development problems. The companies are presented as a force in society that is needed-not just for the creation of employment and economic growth, but also for addressing social problems.
In this way the "truth", constructed in CSR discourse, tends to simplify social problems and insulate CSR from critical perspectives on the role of the company (see for instance, Blowfield & Frynas, 2005;Fleming & Jones, 2013). With the construction of this type of discourse, little attention is paid to the tensions that may arise between a profit motive and the greater good of society; complexities are downplayed and alternative constructions of social issues that might be less beneficial to the company are actively kept outside the discourse on CSR. For instance, constructing CSR as risk management may not necessarily mean that issues related to company production are dealt with in a manner that is the most beneficial to employees, local communities, or the countries where the companies operate. Moreover, issues such as land use or tax avoidance, require alternative constructions of the role of companies in society that take into consideration concepts such as social rights, social justice and inequalities. But these concepts are left out in a discourse that is constructed around the business motive.
The "truths" that the interpretative repertoires convey restrict the boundaries of what can meaningfully be addressed as CSR issues to those that also (financially) benefit the company. The problem is then that the discourse on CSR becomes limited, and limiting, as a way of addressing social issues (see Blowfield & Frynas, 2005;Fleming & Jones, 2013;Valentin & Murillo, 2011). Hence, the social significance of these constructions lies in the fact that they legitimate the power of companies. In doing so, they give a convincing picture of the "good company" that wards off criticism. The discursive construction of the company as part of the solution becomes "truth", institutionalised under the banner of CSR, and shelters companies from criticism.

Differences in discursive constructions: discourse as flexible resources
Across the four country contexts there were at the same time apparent differences in how the companies discursively construct CSR, since the meanings assigned to CSR are context-bound. In the Swedish, UK and South African reporting, all four interpretative repertoires were central in the construction of CSR: asserting the business motive, making a difference, the win-win situation, and being part of the solution. In the South African CSR reporting the win-win discourse was however less encompassing since companies also reported on issues not derived from own ideas on how to merge the profit motive with making a difference.
The Mauritian CSR reporting drew only on the repertoires of making a difference and being part of the solution; since CSR is mandated, there is no need to assert the business rationale or emphasise the win-win logic. In other words, they have no need to justify their involvement in CSR and could therefore construct CSR involvement as focused on making a difference without the need to emphasise that this is congruent with the business logic.
The interpretative repertoires also differed in the way that South African and Mauritian companies focused more on their role within the country, whereas the Swedish and UK-based companies focused on their operations in locations around the world. Discursively constructing a company as part of the solution to social problems requires an orientation towards those problems that can be regarded by the company's stakeholders as legitimate concerns for the company, given the context. Thus, in the context of emerging economies it is relevant for a company to construct CSR largely in terms of responding to local social issues and socio-economic development needs and in this way show a commitment to contribute towards filling the gaps in welfare provision. In the Swedish and UK company reporting other discursive constructions were visible with CSR being linked to company core activities to a larger extent and also connected to location of company operations around the globe. Thus, the empirical findings show that the context influences the CSR approach within the texts in terms of the emerging economy context requiring a different CSR orientation than that of advanced welfare states.
There were few differences in the use of the four interpretative repertoires between Sweden and the UK. Companies in both contexts have adopted very similar discourses on CSR, which may be indicative of how multinational companies based in the West draw on a homogenised global discourse on CSR. While it could be that companies in emerging economies, such as South Africa and Mauritius, may come to follow this approach to CSR it is not evident that this will be the case. What can be seen in South Africa and Mauritius is instead another approach being institutionalised that entails constructions of company CSR as integral to enhancing and providing welfare within the country. As the insights from discursive psychology tell us, discourse is interactionally and contextually constructed: "Since talk and text are action orientated, versions are likely to show variability according to the different interactional contexts they are constructed to serve" (Edwards & Potter, 1992, p. 28). Hence, the interpretative repertoires can be seen as flexible resources that the companies adapt to the context.

Conclusion
The empirical findings illustrate how companies use discourse to construct CSR and how the self-reporting communicates approaches to CSR that legitimate corporate power. By examining interpretative repertoires the article contributes to the debate on CSR with a sharpened analysis of how power and knowledge intersect as companies through self-reporting construct the understandings and accepted "truths" of CSR that come to prevail. The focus on interpretative repertoires has drawn attention to how discourses are actively used as flexible resources in company CSR reporting in order to accomplish social actions (providing legitimacy) within specific contexts (see Jørgensen & Phillips, 2010). In this study the use of discourse was related to national contexts and the empirical findings show how the companies give meaning to CSR in relation to the national context. Thus, the contribution of this article also lies in its illustration of the relevance of the national context for how CSR is discursively constructed. As the empirical findings have shown, how companies construct and uphold the meaning of CSR is, at least in part, shaped by the national context.
Beyond the scope of this article is the extent to which CSR actually contributes to making a difference. A central concern in self-reporting, obviously, is to enhance the company's image. An important point, then, as underscored in this article, is that company constructions of CSR do "communicative actions" (Edwards & Potter, 1992) that serve to legitimate company actions and power, and to ward off criticism. As sites where power is expressed, CSR self-reporting ensures that CSR "addresses only what business is prepared to accept as negotiable" (Blowfield, 2005, p. 523). In producing and circumscribing knowledge on CSR, companies can package CSR in formats suitable to their business purposes. Acknowledging how company power in this way intersects with discourses on CSR is an essential starting point for any discussion of what CSR can offer society.

Implications for future research
The implications of this research for future research are twofold: Firstly, analysing CSR has to be seen as contextual endeavour, and differences in discursive constructions across contexts are relevant to examine further. In future studies of differences in discursive constructions between emerging and established economies temporal analysis of how CSR has evolved could be useful. Other contextual aspects of CSR can also be further examined, such as company size, since the approaches of smaller companies to CSR may differ much from those of large multinational companies such as the companies analysed in this study. Secondly, CSR self-reporting is an image-making enterprise, and the constructions of CSR that have emerged are self-serving, which cannot be overlooked if CSR is to be assessed in terms of its contributions. It may be argued that the self-interest and profit motives are useful means to engage companies to contribute towards the more desirable end of addressing social problems and meeting social needs. However, doing so would be to disregard how power intersects with knowledge production on CSR and how constructions of CSR are constitutive of social realities: the meanings given to CSR set the direction for social actions. As companies construct CSR, they also contribute to producing discourses and "truths" about company engagement in social issues; they contribute to the institutionalisation of CSR's meanings. The interpretative repertoires identified in the analysis are examples of how this is done. These implications suggest a need to link company CSR reporting with perspectives from various other groups in society commenting on them, including trade unions, NGOs and community-based organisations, and in this way shed light on alternative constructions of CSR that may be omitted from the institutionalisation of the concept.