Communicative legitimacy through CSR:
The Interplay Between Corporation, Media and Protest Actors
Carla Donayre, 10700846
Master Thesis
Graduate School of Communication
Master’s programme Communication Science
Supervisor: Dr. Friederike Schultz
June 2015
Abstract
This research contributes to the study of the interplay of corporate communication, news media and protest actors based on a constructivist perspective of communication and corporate social responsibility (CSR). Drawing on a communicative view of CSR and legitimacy, the study overcomes the limitations of instrumental perspectives and expands CSR communication as an emerging social process where multilevel actors negotiate
meanings to reproduce and shape organizations; which influences the organizational identity. By presenting the notions of associative framing and network agenda-building, the study enlarges agenda-setting and –building concepts and proposes a more complex understanding of the relation between actors and their discourses, in which meanings emerge as multiple and non-unidirectional connections. Based on a semantic network and an automatic content analysis, the study analysis more than 3600 documents of the last ten years (2005-2014) corresponding to Shell, UK media and UK protest groups to compare first-level agenda-setting effects and frames. Results reveal that Shell has institutionalized CSR around very specific dimensions, with great emphasis in environmental and philanthropic responsibility, and that it maintained very complex discourses in terms of number and strength of issues, actors and associative framings over time. Media, on the other hand, presented simple and atomized discourses, while protest actors varied over time.
Keywords: Corporate social responsibility, communicative legitimacy, agenda-setting and -building, interplay of actors, media, protest actors, organizational identity, semantic network analysis, AmCAT.
Communicative legitimacy through CSR:
The Interplay Between Corporation, Media and Protest Actors
Corporate social responsibility (CSR) has increased in attractiveness in the last decades for number of corporations finding in the concept a solution to align their values, practices and expectations with those from society. At the same time, more public pressures arise from different fronts raising the bar not only in terms of innovation and efficiency needs, but also of normative standards and expectations to behave in certain ways.
Organizations and researchers have traditionally discussed CSR from managerial views as a tool to control and manipulate stakeholders to fulfill social expectations and gain better reputation. Conventional research in CSR communication (Bronn, 2010) supports this idea based on the belief that information goes from messengers to receivers as packages, with no misunderstandings occurring, reducing communication to a symbolic manipulation or to a guarantee for consensus.
However, the multiplicity of voices that interact in reality demands a broader understanding of what corporate communication and CSR are. Studies on the constitutive perspective of communication (Christensen & Cornelissen, 2011; Schoeneborn and Trittin, 2013; Schultz, Castelló & Morsing, 2013; Schultz & Wehmeier, 2010) broaden the notion of CSR as an emerging, conflictive and disintegrative communicative event through which legitimacy is negotiated and built. Legitimacy is, therefore, communicatively constructed (Schultz et al., 2013) in a process where multilevel agents individually make sense of events and, simultaneously, interact and negotiate meanings.
The interaction of actors has been observed in multiple studies on public relations, news media and publics, for example, regarding crisis situations (Schultz, Kleinnijenhuis, Oegema, Utz, & van Atteveldt, 2012). In general, research observing the relation between media, corporations and public are focused on the transfer of information through lineal and logical dynamics and for relatively short periods of time. Yet, few studies analyze these
interactions in a large time frame that would allow the development and institutionalization of CSR, and its effects on the organizational identity.
Based on a longitudinal case study (ten years) developed via an automated content analysis, the present thesis aims at partially filling this gap, guided by the following main research question: How is CSR communicatively (de)constructed and institutionalized in order to build legitimacy in the dynamic interplay between corporations, media and protest actors discourses? To the different aspects implied here, we propose a model that, first, asks about the implications of corporate CSR discourses in the process of organizational identity construction, leading to the first specific research question: How does the communicative (de)construction of CSR change over time within the identity process? Then, in order to observe the interplay of actors, we focus on the semantic and social networks (van Atteveldt, 2008; Schultz et al., 2012) to explore the notion of network agenda-setting and –building and, consequently, expand traditional approaches that underestimate complex associations in discourses. This observation tries to respond to other three research questions: RQ2: How are associative frames in the interplay between corporation and media discourses related with each other?; RQ3: How are associative frames in the interplay between media and protest actors discourses related with each other?; and RQ4: How are associative frames in the interplay between protest actors and corporate discourses related with each other?
The societal relevance of our study lies in its empirical contribution to a deeper comprehension from organizations (including managers, communication professionals and staff in general) of the collective and chaotic nature of communication, and the awareness of how organizational issues can be interpreted and portrayed by different actors.
The study is organized in four sections. The first one, built on the constitutive
perspective of communication, discusses CSR and legitimacy building and examines agenda-setting and -building concepts as the theoretical basis to observe the interplay between different social actors and the construction of discourses. The second section presents the
operationalization of the study including the research method, research design, units of analysis and data collection. A third part offers the empirical results from the automated content analysis. Finally, the discussion section presents the implications with regards to theory, practical implications and suggestions for future research.
Theoretical framework
CSR and communicative legitimacy: The social construction of CSR
Important literature on management and corporate communication (Brønn, 2010) defines corporate communication as a management function used to coordinate the overall organizational communication in order to guarantee reputational assets (Christensen, Morsing, & Cheney, 2008). Based on a “conduit” metaphor (Axley, 1984), this view guarantees the self-reassurance of senders assuming that communication implies a simple transfer of meanings and the fidelity between sent and received messages (Axley, 1984). Thus, corporate communication is seen as a linear, unidirectional and orchestrated process (Schoeneborn & Trittin, 2013) where miscommunication rarely occur and receivers act as “unpackers” of predetermined messages.
This instrumental approach disregards the social nature of communication and its role in constituting the social order (Craig, 1999; Schoenerborn & Trittin, 2013; Christensen & Cornelissen, 2011). Constructivist approaches argue that through communication actors make sense of the world by (de)constructing meanings (Christensen & Cornelissen, 2011; Hatch & Schultz, 2002), which settles the basis for social coordination (Christensen & Cornelissen, 2011; Schultz et al., 2013; Schoeneborn & Trittin, 2013). Communication is an ongoing social process of interaction where reality emerges (Putman & Nicotera, 2008) and through which it is continuously (de)constructed. As identities are multiple, this reality is not unique, but diverse. Managing communication as a whole is, therefore, pure illusion.
Organizations as communicative events. As social constructs, organizations are not
simple entities where communication occurs, but continuing events enabled by and consisted of communication (Cooren, Kuhn, Cornelissen, & Clark, 2011; Ashcraft et al., 2009). As communication organizes reality, it consequently enacts organizations by constantly reproducing them through interactions. “It is through communicative practices, and not primarily through specific individual human members, that organizations are created and sustained” (Schoeneborn & Trittin, 2013, p. 198).
According to the communication as constitutive of organizations (CCO) perspective, interactions and symbolic negotiations between multiple voices make organizations a
polyphonic, dialogic and chaotic social space. As “actors within and outside the organization tend to engage in discursive struggles on sensegiving and framing” (Schoeneborn & Trittin, 2013, p. 199), internal and external dynamics, and third parties (e.g. external stakeholders), are intrinsic to the constitutive nature of communication (Christensen and Cornelissen, 2011; Morsing & Shultz, 2006; Schultz and Wehmeier, 2010).
Following this reasoning, our main research question is as follows: How is CSR communicatively (de)constructed and institutionalized in order to build legitimacy in the dynamic interplay between corporations, media and protest actors’ discourses?
Communicative legitimacy and CSR. Organizations exist because they are
legitimate, i.e., they are socially accepted and aligned with social norms and expectations. Legitimacy is constructed in and through an ongoing communication process, and can be defined as the perception that the actions of an organization are congruent with a social system of norms and expectations (Palazzo & Scherer, 2006; Ashforth & Gibbs, 1990). It is a “collective making of meaning” (Ashforth & Gibbs, 1990, p. 186), which makes it a complex and problematic process: Multiple and contradictory pressures come from internal and external stakeholders, and simultaneously organizations need to satisfy their own ideals (Lee & Carroll, 2011; Christensen et al., 2013; Palazzo & Richter, 2005). Regarding those ideals,
Lee & Carroll (2011) defined the most relevant dimensions of CSR: Economic (based on the idea that companies must be profitable in order to exist), philanthropic (focused on being a good corporate citizen), legal (focused on the respect for the rules of law), and ethical responsibility (referred to the obligation to do what is right and avoid harm). The environmental dimension was included later.
Instrumental perspectives define CSR programs as stakeholder management tools that guarantee reputation, economic profits, and moral acceptance. Oriented at financial benefits and solving discrepancies, these views refer to pragmatic legitimacy (obtained through the manipulation of symbols or ‘symbolic management’), and cognitive legitimacy (achieved through behavioral adaptations to social expectations or ‘substantive management’) (Ashforth & Gibbs, 1990). Political-normative views focus on the alignment of organizations and their environments (Patriotta, Gond, & Schultz, 2011), and provide moral legitimacy based on deliberative communication (Ihlen, Bartlett, & May, 2011; Schultz et al., 2013). Both views privilege corporate communication and CSR as ways of control, manipulation and consensus (Palazzo & Scherer, 2006; Schultz et al., 2013).
How can legitimacy be maintained over time when it is seen as a specific symbolic reaction to external pressures, is based on ephemeral organizational changes or is built based on an illusionary idea of communicative consensus? These questions raise again the
limitations of functionalist perspectives (corporate communication is a tool of manipulation, constituents are passive actors, legitimacy is a manageable output rather than a constructed process, and communication reaches legitimacy by bringing everybody to an agreement), and at the same time open a space for communicative legitimacy as a more holistic understanding of organizational reality where CSR rather than a corporate program is a communicative dynamic of meaning negotiation in which multiple expressions interact.
The definition of CSR is, thus, oriented towards an emerging communicative event of dissonant and contradictory voices that make sense of events (Schultz et al., 2013). CSR must
deal with “the contradictions and paradoxes that may arise between different motives for practicing CSR” and balance “the ideal of corporate citizenship and demands for
profitability” (Schoeneborn & Trittin, 2013, p. 201). It implies misalignments, mistrusts and break-ups as necessary conditions for legitimacy building. CSR communicatively builds legitimacy through a joint sensemaking and “consensus-building” based on the continuous struggle between multiple actors (Patriotta et al., 2011).
Communicative legitimacy recognizes that audiences and constituents are active agents constructing multiple justifications through rhetorical strategies for legitimacy
building (Patriotta et al., 2011). Patriotta et al. (2011) highlighted the interplay between actor and the importance of balancing their logics “so as to develop justifications robust enough to withstand public scrutiny” (p. 1828). In other words, the interplay must not reach an endpoint of polarization that breaks-up communication, but should get to a point of synchronization where certain interpretations about objects of reality and their associations are shared.
CSR as institutional work: Between institutionalization and sensemaking.
Institutional work is fruitful at this point as it offers a holistic view that integrates institutional and sensemaking perspectives for the understanding of legitimacy (Lawrence, Leca, &
Zilber, 2009). CSR institutionalization can be seen as a form of institutional work where macro (environment and institutions), meso (publics and audiences) and micro (internal constituents) agents interplay in the production, maintenance and transformation of institutions (Schultz & Wehmeir, 2010).
Institutional work integrates i) the idea that legitimacy is built on the organization’s respect for norms and rules (Schultz & Wehmeir, 2010, p. 11), and that institutionalization of practices such as CSR results from interactions between organizations and external actors (institutional view), with ii) the importance of individuals interpreting symbols (Weick, 1988) (sensemaking) and providing “an account […] that has a semantic function as an
a basis to influence and shape the interpretations of others and negotiate a common understanding with them” (Cornelissen, Carroll, & Elving, 2009, p. 3). This integration grants individuals and collectives a reflexive and intentional role (Lawrence et al., 2009).
As institutional work, CSR is a multilevel meaning negotiation or a “messy”
translation of actors interplaying and reproducing, adjusting or replacing institutions (Schultz & Wehmeir, 2010). The institutionalization of CSR implies a dichotomy of what
organizations say and do (one of the main critics to CSR). However, the distinction between discourse and action is not entirely clear since “communication and action are intimately linked in all process of organizing, because saying is doing and because actions inevitably ‘speak’” (Christensen et al., 2013, p. 5). So, “certain forms of talk can be action and certain speech acts (such as promises) can talk the communicative reality they speak of into being” (Schoeneborn & Trittin, 2013, p. 194). Hence, CSR communication is an “aspirational talk” with performative consequences: Even when CSR discourses are decoupled from real actions, CSR talk states promises that may lead to organizational changes (Scheneborn & Trittin, 2013). “A CSR program may, in this view, be a first step toward a desired future” (Christensen et al., 2013, p. 5). Since saying something is performing an act (‘performative potential of communication’) (Christensen et al., 2013), sooner or later CSR communication unchains actions that confirm those proclamations and reaffirm organizations as legitimate actors (Schultz et al., 2013; Palazzo & Ritcher, 2005).
CSR and the (de)construction of organizational identity. CSR communication works as identity claims of both CSR practices and promises about a desired future. Hatch and Shultz (2002) defined organizational identity as a set of processes through which cultural self-understandings and images of internal and external actors continuously cycle within and between each other. Based on a culture-identity-image model, images of the organization affect the internal ones and impact the organizational identity. The organizational culture also affects identity by providing cues for the interpretation of images. For externalizing their
cultural understandings, organizations talk about themselves by incorporating their
reflections into discourses through identity claims, e.g. CSR statements. Finally, identity is impressed on others through the projection of organizational images and the influence of other uncontrolled sources, such as media or protest actors.
Thus, CSR communication is a way of operating and reaffirming the organizational identity, and a form of self-description primarily designed for external publics but
“transformed into ideal self-enhancing and self-confirming images” (Christensen, 1997, p. 203). This makes it a system of auto-communication (Christensen, 1997); i.e. a set of self-referential practices through which organizations reaffirm and transform their images, values and identity. This theoretical argumentation leads to a first research question:
RQ1: How does the communicative (de)construction of CSR change over time within the identity building process?
Construction of legitimacy in the interplay of actors
Individuals not only learn from media about specific issues or events, but also shape their images about them based on the amount and type of emphasis media put on those topics. McCombs and Shaw (1972) proposed the notion of agenda-setting to study the influence of media on publics in the field of political communication defining it as the ability of media to transfer the salience of objects from their agenda to the publics agenda.
Later, its application was extended to study broader objects, such as corporate actors and even nations (Carroll & McCombs, 2003; Kiousis, Popescu, & Mitrook, 2007; Meijer & Kleinnijenhuis, 2006). Based on agenda-setting, the notion of agenda-building was developed to observe the influence of other agendas (e.g. PR) on the formation of the media agenda (Kiousis, Mitrook, Wu, & Seltzer, 2007). In line with previous studies (for example Schultz et al., 2011), agenda-setting and -building are discussed to observe the interplay between corporation, media and protest actors narratives.
The interplay of organization and media: Agenda-building. Agenda-building is a broader concept of agenda-setting proposing the reciprocal influence of different actors agendas (e.g. corporations) in the definition of the media agenda (Ohl, Pincus, Rimmer, & Harrinson, 1995, p. 91). Studies on agenda-building often focus on the influence of PR activities on the media agenda (Kiousis et al., 2007). Indeed, empirical findings confirmed a positive relation between the salience of issues in PR and the attention of those issues in the media, and the influence of PR practices on journalists’ discourses (Parmelee, 2013; Kiousis et al., 2007). Studies (Ihlen & Pallas, 2014) also propose the notion of ‘mediatization’ to highlight the influence of media on how actors relate with and understand each other. Organizations intentionally engage in mediatization through PR in order to reach stakeholders. Media are one of the most important, so organizations try to get the media attention and endorsement to reinforce their legitimacy (Ihlen & Pallas, 2014).
Media, on their side, exert a role in shaping organization discourses and images by building “normative, regulative, and cognitive bases on which corporations are evaluated both as individual organizations as well as societal institutions” (Ihlen & Pallas, 2014, p. 7). They also influence organizations by moving issues from one field to another and shaping the public debate (Pattriota et al., 2011). The relationship is of mutual influence.
As in agenda-setting, in agenda-building the influence between agendas is produced at different levels. Level one (salience of objects) and level two (salience of attributes) were introduced as the initial stages of the agenda-setting studies (Kiousis et al., 2007). Objects are defined as anything that a person has an opinion about (Guo, Vu, & McCombs, 2012), i.e. issues and actors. Agenda refers to the issues or actors where the emphasis is put on.
First-order agenda-building is about the transmission of object salience based on the amount and frequency of coverage. Salience refers to the visibility of the object; so, it is expected that the more visibility certain objects have in the corporate agenda, the more awareness about them in the media. By defining an agenda of objects, an organization
suggests what to think about (McCombs, Llamas, Lopez-Escobar, & Rey, 1997). As seen, CSR communication aims at presenting images about corporate identity and practices that portray corporations as social responsible actors. Based on this logic, hypothesis 1 is as follows:
H1: The increase of salience of CSR objects in corporate discourses leads to an increase of coverage of those objects in media discourses.
The second level offers an agenda of attributes based on the frequency of appearance of attributes (Guo et al., 2012). These attributes configure the evaluations of those objects, and shape the comprehension about them (Carroll & McCombs, 2003; McCombs et al., 1998; Schultz et al., 2012). Cognitive and affective dimensions based on an agenda of substantive attributes and an agenda of affective attributes (Carroll & McCombs, 2003; Guo et al., 2012; Kiousis et al., 2007) are in play. Media tells us “how to think about some objects” (McCombs et al., 1998, p. 704). For this study, only first-level agenda-setting and –building and
associative framing will be developed since our main objective is to explore the attention for objects and the complex connections between them rather than their specific attributes.
Traditional views of agenda-setting and –building assume that individual’s information processing respond to a logical and linear mental dynamic (Guo et al., 2012). This, however, limits the transfer of salience as discrete processes where objects and their features are ranked in an ordered list. ‘Third level’ agenda-setting or ‘network agenda-setting model’ is an emerging concept (Schultz et al., 2012; Guo et al., 2012) developed to overcome this limitation. It distinguishes non-hierarchical relations between issues, actors and
evaluations of any other concepts related to them.
Third level agenda-setting relies on the idea that “individuals’ cognitive
representation of object and attributes is presented as a network-like structure where any particular node will be connected to numerous other nodes” (Guo et al., 2012, p. 54). The transfer of salience of diverse agendas is, therefore, a simultaneous process where elements
are interconnected. Strategic framing is about “not only the strategic selection of concepts such as issues or actors (agenda) and their evaluation, but also of their mutual associations” (associative framing) (Schultz et al., 2012, p. 3). Therefore, associative frames transcend the bilateral connections proposed by first and second levels, and propose simultaneous, multiple and non-unidirectional connections to observe frames as “semantic networks”. Meanings emerge from those multiple associative frames, and from the interaction between them and existent associative networks (Guo et al., 2012). This means that the construction and reconstruction of associative frames is based on the connection between new and old information that can create new nodes or modify the existing ones.
As third-order agenda-building allows a deeper insight on the connections between different agendas and their concepts (issues, actors and attributions), our second research question aims at responding how associative frames in corporate and news media narratives are interconnected:
RQ2: How are associative frames in the interplay between corporation and media discourses related with each other?
The interplay of media and protest actors: Agenda-setting. Media not only shape the publics attention and evaluations about an organization, but can also transform a
corporate issue into a social problem of public concern. Indeed, a longitudinal study of 25 years by Lee and Carroll (2011) showed how the agenda-setting role shaped the evolution of CSR as a public issue and how it changed in relation with the media’s interests and
expectations. In media, cues such as the amount and frequency of exposure, the size of the article and headline, or the type and number of pages designated to cover stories have a direct relationship with the audiences perception about those objects and with the place they occupy in the public’s agenda (Carroll & McCombs, 2003). Accordingly, it is expected that the more visibility news media give to certain CSR issues and actors, the more aware protest actors will be about them, which leads us to hypothesis 2:
H2: The increase of salience of CSR objects in media discourses leads to an increase of awareness about those objects among protest actors.
It is also of interest to observe the reciprocal and non-directional relationship between media and protest actors discourses, which is guided by our next research question:
RQ3: How are associative frames in the interplay between media and protest actors’ discourses related with each other?
The interplay of protest actors and organization: Stakeholder agenda-setting. From a bourdieuian perspective, PR can be understood as a struggle of power in which corporations are expected to influence other actors discourses (Ihlen, 2007; Edwards, 2012). Indeed, empirical findings on political communication have showed positive connections between PR content and public opinion (Kiousis et al., 2007), and confirmed the influence of PR on stakeholders via the transfer of salience of objects from the corporate to the publics agenda (Kiousis et al., 2007).
However, since “power over agendas might arise from ‘networks of relationships and patterns of interaction’” (Hunter, Van Wassenhove, Besiou, & van Halderen, 2013, p. 28) between different actors, stakeholders can also shape others’ attention and perceptions based on their own communication strategies. Stakeholder agenda-setting observes this influence (Hunter et al., 2013). Hunter et al. (2013) proposed ‘stakeholder media’ as the media used by stakeholders to influence corporations and mobilize communities of support by setting the public agenda and by framing corporate practices. Traditionally reserved to the media, the ‘gatekeeper’ role has been extended to a stakeholder function and, therefore, stakeholders decide what and how to publish issues and actors.
As in agenda-setting and –building, first level stakeholder agenda-setting regards the transfer of salience of issues and actors. Thus, hypothesis 3 is as follows:
H3: The increase of salience of CSR objects in protest actors discourses leads to an increase of salience of those objects in corporate discourses.
As for associative frames, the idea is to observe the possible mutual associations resulting from the interplay of stakeholder groups and the corporation guided by following research question:
RQ4: How are CSR associative frames in the interplay between protest actors and corporation discourses related with each other?
Operationalization of the Study Research Design
A longitudinal case study was carried out in order to respond to the proposed research
questions and hypotheses focused on the (de)construction of Royal Dutch Shell CSR within a period of ten years in the interplay with news media and protest groups discourses. From a more theoretical perspective, we aim at observing how Shell CSR communication works as self-descriptions that lead to the institutionalization of CSR practices and the legitimation of the organization within the social system where it operates.
Shell is an Anglo-Dutch multinational company with headquarters in The Netherlands and a central office in the United Kingdom. It is important for being one of the biggest corporations in the extractive business industry. We took Shell as it continuously attracts
Corporate CSR discourses - Attention - Associations Media discourses - Attention - Associations
Protest actors discourses - Attention - Associations H1 Corporate identity RQ1 H3 H2 RQ2 RQ4 RQ3 Figure'1.'Conceptual+model++
large amount of media and protest groups attention. From some years now, Shell has developed a CSR discourse depicting itself as an environmentally friendly corporation respectful of human rights and social justice. This discourse, however, clashes with others’ perceptions about its performance as a fair social actor. The long time-interval analyzed which allows us to observe patterns and changes in the different discourses over time and to get more concise results that contribute to the validity of the study.
Units of analysis and data collection
Corporate reports, newspaper articles, and protest actors press releases integrate the corps of analysis (see Table 1 for the data overview). Each single material is a unit of analysis.
Corporation - Royal Dutch Shell. Corporate communication was analyzed based on ten Annual Sustainability Reports issued by Shell between 2005 and 2014. Reports from 2007 to 2014 were retrieved from its corporate website the 23rd of April 2015, and those from
2005 and 2006 from the NIOC Information Center and Central Library
(www.nioclibrary.ir/oil-companies.htm) the 24th of April 2015. Annual reports were not considered as are mainly focused on financial performance, which could have biased the analysis. The data correspond to the global office, as the study focuses on the global
corporate CSR and not on one country’s office. Due to practical requirements of the software used for the analysis, each report was split in different sections using similar criteria.
News media. The collection of newspapers articles issued by The Guardian and The Financial Times was done during the first week of May 2015. Articles from the 1st of January 2005 until the 31st of December 2014 were collected via LexisNexis through a search query
for: (Shell AND social responsibility) AND (Shell AND sustainability) AND (Shell AND oil spill) AND (Shell AND climate change) AND (Shell AND environment) AND (Shell AND pollution) AND (Shell AND Nigeria). We focused on UK media, as the company is one of the most important in the UK with a primary listing on the London Stock Exchange (the largest in 2013). Dutch newspapers were not included due to language limitations. The
newspapers were selected according to (1) the representativeness of the UK different media discourses and (2) the availability of English articles in the online source LexisNexis. Regarding their representativeness, we chose one newspaper from centered-left and social liberal democratic views (The Guardian), and one economically liberal and politically centrist with conservative views (The Financial Times).
Protest actors. Press releases, articles, blogs, statements and reports issued by Greenpeace, Friends of the Earth, Platform UK and Amnesty International from January 2005 to December 2014 were collected from their official web sites between the 26th of April
and the 5th of May 2015. The four organizations were selected for their importance among international non-profit organizations and for being highly engaged in the monitoring of Shell activities. All of them have operations in the UK.
Methods of analysis
This longitudinal case study was developed via a content analysis relying on qualitative and quantitative techniques to explore attention on issues and actors, and associations between them within a period of 10 years. Content analysis was performed through a semantic network analysis of 209 corporate documents, 3373 news articles and 326 NGO press releases and blog. The analysis was developed in two stages:
Table 1.
Overview of the data collection presented by source
Corporation* Media** Protest actors***
2005 19 264 13 2006 22 371 16 2007 19 362 16 2008 21 381 20 2009 20 305 56 2010 20 399 38 2011 19 236 50 2012 21 386 46 2013 27 332 31 2014 21 337 40 Total 209 3373 326
* Royal Dutch Shell
** The Guardian, The Financial Times
Preliminary exploratory analysis. Before starting with the semantic network analysis, we carried out an exploratory analysis in order to elaborate a list containing the main issues related to Shell CSR between 2005 and 2014. Based on literature and on a review of all the corporate documents and random selected protest actor’s documents, a keyword list was elaborated with important descriptors referred to Shell CSR. The first keyword list was adjusted by performing a simple frequency analysis to verify the most mentioned descriptors and do a category grouping. Drawing on Schultz et al. (2011) procedure, a validity test was performed to filter first-degree mistakes (“text elements wrongly taken as a hit”) and second-degree mistakes (“text elements wrongly skipped as a hit”). Appendix A presents the final keyword list with the key actors and issues. The seven actors are the most mentioned by the company and by protest actors over the ten-years period. The eleven categories of issues were defined based on the CSR dimensions defined by Lee and Carroll (2011) and on concepts related to reputation and legitimacy. An additional frequency analysis was carried out to observe more in detail the attention that the organization, the media and the protest actors put on those actors and issues.
Semantic network analysis. The analysis of changes and associations was performed via an automated content analysis via AmCAT (Amsterdam Content Analysis Toolkit). AmCAT is a document management system (navigator and database) used to perform large-scale content analysis through a keyword-based analysis. It allows identifying networks of particular objects extracted from the materials or linked to other more abstract issues or actor categories (van Atteveld, 2008). We used the tool to measure attentionon actors and issues (first level agendabuilding), and the associations of any concept related to them (third level agenda building, see Schultz et al., 2012).
In AmCAT messages are represented as a network of objects; i.e. network of relations between actors (van Atteveld, 2008), based on asymmetric conditional probabilities resulting from occurrences and co-occurrences of concepts (p formula) per unit (units of time).
Associative framing (or conditional probabilities) is portrayed by a diagram in which actors and issues are connected via arrows of different thicknesses representing the strength of the conditional probability. An arrow goes from concept a to concept b if a is framed in relation of b, i.e. “if the conditional probability that b will discussed once a is discussed if very high” (Schultz et al., 2012, p. 101). For a better visualization of the network diagrams we left aside Shell and defined different cutoff levels for the different narratives.
Results
Agenda-building and associative framing in Shell public relations. Attention for issues and actors.
Overall attention. For the ten years time, Shell gave very high attention to four dimensions of CSR (see Table 2): Environmental responsibility—which plays the most relevant role (100%)—, economic responsibility (92%), philanthropic responsibility (89%) and legal responsibility (80%). Ethical responsibility, contrarily, was less present (55%), even less than reputation and legitimacy (66%), showing the relative low interest of the
corporation in moral claims. Without underestimating the economic aspect, Shell clearly prioritized environmental responsibility as the basis for its business success: “[…] our short- and long-term business success depends on finding environmentally and socially responsible ways to help meet the world’s future energy needs” (Shell, 2005a). The high attention to philanthropic responsibility shows that Shell also emphasized being a good corporate citizen by going beyond its core business and contributing to the community (voluntary work, prevention of malaria and HIV/AIDS, activities through the Shell Foundation). “We aim to be a good neighbour to these communities” (Shell, 2009).
When comparing to the news media agenda, it can be observed that environmental responsibility received less attention in newspapers (64%) than in corporate reports and was not the most important topic. Contrarily to Shell, media gave more attention to legal (50%)
than to philanthropic responsibility (32%) prioritizing the role of corporations in meeting social expectations established by law. Also, although reputational and legitimacy had a high role in Shell reports, the attention was not transferred to the media agenda where the issue was much less mentioned (16%) and slightly overtaken by negative reputation (17%).
With regards to actors, we can see that affected populations (71%) have the most dominant role in Shell discourses. Actually, they are constantly mentioned with positive connotations as target populations benefited from Shell’s actions: “Indigenous communities can benefit from oil and gas projects, but there is a need to protect their traditional lifestyles (Shell, 2009). Partners (53%)—many of them environmentally oriented, e.g. Earthwatch, the UN Global Compact or Wetlands International—were also important actors that help Shell validate its principles and standards. When comparing Shell and media agenda, media granted more predominant roles to companies similar to Shell (52%) and to governmental actors (36%). Since both actors are important economic actors, it can be assumed that the high attention they received was linked to the importance of economic responsibility.
Attention over time. To observe changes in attention, Appendix B gives a complete overview for the three periods. The use of CSR as a general concept to describe its CSR
Table 2.
Overview actors and issues
Articles Hits % Articles Hits % Articles Hits %
Actors Shell 209 5348 100 3110 11198 90 325 4887 100 Government actors 65 202 31 1253 3340 36 180 677 55 Protest actors 89 170 43 978 2666 28 260 1504 80 Affected populations 148 669 71 1022 1833 30 152 386 47 International organizations 71 147 34 890 2135 26 130 321 40 Similar companies 61 177 29 1777 8709 52 100 458 31 Shell partners 110 509 53 560 830 16 30 112 9 Armed groups 4 6 2 222 511 6 49 244 15 Issues CSR 56 88 27 141 296 4 30 60 9 Economic responsibility 192 1300 92 2393 8227 70 148 374 45 Legal responsibility 167 408 80 1732 3884 50 221 775 68 Ethical responsibility 114 497 55 427 556 12 55 114 17 Environmental responsibility 209 3683 100 2196 8750 64 283 2202 87 Philantropic responsibility 186 2138 89 1111 2385 32 210 1041 64 Reputation/Legitimacy 138 526 66 565 810 16 69 125 21 Negative reputation 14 21 7 586 775 17 181 446 56 Negative impact 80 126 38 270 362 8 107 313 33 Reaction 17 19 8 422 564 12 133 274 41 Other crises 9 11 4 223 396 6 24 42 7
N (# items with Shell included) 209 3372 326
Protest actors News media
activities decreased over time (32%, 24%, 23%). Contrarily, it has constantly focused on the environmental dimension as the main aspect, and in a smaller degree on the philanthropic, economic and legal ones. Although environmental responsibility has occupied the main role over time (100% in the three phases), it is interesting how in 2005-2008 economic
responsibility (99%) competed with it and had practically the same level of importance. From 2009, however, the economic dimension started losing strength for Shell (93% in 2009-2011 and 83% in 2014), which gave the philanthropic aspect more relevance (87% in 2012-2014). This last dimension, and the legal one, has been stable overtime, contrarily to ethical responsibility, having important attention in the second period (53%, 61%, 51%), showing that moral views were not prioritized over other dimensions. Interestingly, crises from other companies appeared in Shell discourse only since 2009-2011 (8%) due to the BP oil spill of 2010. This could also explain the increase of attention for similar companies from 2005-2008 to 2009-2011 (25% to 32%) and for affected populations (68% to 75%). In the words of Shell: “The BP Deepwater Horizon […] will have repercussions for many years to come. The incident became an environmental disaster that affected communities […] It will take time for our industry to recover credibility (Shell, 2010).
To test hypothesis 1, proposing that the increase of salience of objects in corporate discourses leads to an increase of coverage of those objects in media discourses, we
compared the evolution of both frames. Coincidences were focused on the predominant role of economic and environmental responsibility over time. It is interesting to see, however, how the attention for environmental responsibility did not vary in for Shell from one period to another, but it did in media. The BP crisis could explain why media reduced the attention for economic aspects between 2005-2008 and 2009-2011 (75% to 67%) and increased it for environmental (60% to 69%), legal responsibility (46%, 52%) and negative reputation (13% to 20%). Albeit the attention Shell gave to CSR, the insignificant attention in media (6%, 3%, 3%) also confirmed their interest for specific dimensions. With regards to actors, Appendix B
shows that the highest increase of attention in media was for similar companies in the period 2009-2010 (52%, 60%, 47%), when the industry was affected by Gulf of Mexico oil spill. Inversely to Shell agenda, during the same period the attention for governmental actors increased in newspapers (33% to 40%), which could suggest that media were asking for the intervention of governmental agents in the oil industry. For Shell, the role of government actors was more relevant in recent years (passing from 25% in 2009-2011 to 36% in 2012-2014), where partners and legitimacy aspects were also highlighted. “Renewables […] will continue to grow with strong government support […] More government funding [is] needed to move forward” (Shell, 2012).
Since similarities in the fluctuation of some CSR in both discourses were identified in the first two phases, but divergent frames appeared in the last period, an agenda-setting effect was partially produced. Thus, hypothesis 1 is partly accepted.
Associative framing.
Overall associations. Complete results of the semantic network analysis and a diagram with the main relationships are shown in Appendix C and G.Overall, strong associations between environmental responsibility and multiple concepts are observed, confirming the central role of the environmental dimension in Shell discourse. Among the strongest relations are: The one with government actors (p(EnvironmResp|GovActors) = 0.98), with other crises (p(EnvironmResp|OtherCrises) = 0.98), and with similar companies (p(EnvironmResp|SimCompanies) = 0.98). As Shell suggested, environmental responsibility requires the engagement of governments and other companies: “Climate change poses the biggest challenge […] man-made climate change is underway [and] it will require action on many fronts […]. Change will likely come from the combination of government policies and new technology, developed and rolled out by companies” (Shell, 2006).
As for frames (Appendix G), it can be seen that Shell semantic network is much more complex than the media frame as many strong associations connect important concepts. For
example, environmental, philanthropic and economic responsibility are strongly associated to all the other concepts and also between them (minimum associations of 0.77). Contrarily, media associative framing showed a discourse based in two main clusters: The environmental one, moderately associated with negative reputation, protest actors and international
organizations, and, the economic frame, related to reputation and legitimacy topics, and to CSR. The main difference with Shell is not only that fewer concepts were linked to the main dimensions, but also that each concept’s correlation was weaker (e.g. maximum association of 0.70 in media versus associations above 0.92 in Shell for the environmental issue).
Also, Shell highly associated itself even with those concepts retrieved from protest actors documents that could have a negative connotation, such as negative impact
(p(Shell|NegatImpact) = 0.99). However, it used them as examples of what the company avoids doing or situations that pretended to change (e.g. the zero fatalities policy or the prevention of deaths by launching the Global Alliance for Clean Cookstoves).Media portrayed Shell in relation to similar companies (p(Shell|SimCompanies) = 0.77) rather than in terms of environmental responsibility (p(Shell|EnvironmResp) = 0.66).
Associations over time. Appendix D shows the conditional probabilities in Shell discourses for the three periods. Results confirm the high complexity of Shell discourses as many concepts are highly related. The key role of environmental responsibility is seen over the ten years. During 2005-2008 Shell has largely focused on the connections between environmental, economic and philanthropic responsibility with all issues and actors (minimum correlations of 0.92, 0.90 and 0.72, respectively). In 2009-2011 the discourse slightly turned towards philanthropic responsibility through the strengthening of its relationships with other concepts—always keeping the environmental dimension in the center—and the weakening of the links of the economic aspect (the Business Skills for World Heritage programme launched in 2009 and the support to the Global Alliance for Clean Cookstoves in 2010 illustrate this interest). Also, two actors gained presence in Shell speech:
Affected populations, linked to negative reputation (p(AffectPopulat|NegatReput) = 1) and to environmental responsibility (p(EnvironmResp|AffectPopulat) = 0.96); and partners, related with environmental responsibility as before (p(EnvironmResp|Partners) = 0.99) and with negative reputation (p(Partners|NegatReput) = 0.98). Connections are illustrated here: “Security issues and sabotage are constant threats […], but we are cautiously optimistic that conditions are improving and that our initiatives to help community development and build on our partnerships will produce good results” (Shell, 2010). As legal issues arose in the energy sector, logically the link between legal and environmental responsibility
(p(EnvironmResp|LegalResp) = 0.97) was strengthened. During 2012-2014 Shell brought all concepts towards environmental responsibility making the discourse less dispersed.
To respond to research question 2 (“How are associative frames in the interplay between corporation and media discourses related with each other?”), Shell and media frames were compared over time. In Appendix G and H we see a much higher complexity in Shell discourses than media. Shell associative framings in every period include many more concepts associated between them and stronger correlations between them. Indeed, in the first period media focused on economic responsibility to which moderately related three others: Partners, reputation and CSR (p(EconomResp|Partners) = 0.68, p(EconomResp|ReputLegit) = 0.70, p(EconomResp|CSR) = 0.83). In 2009-2011, while Shell discourse became more
complex, media speech was dichotomized into two topics: Similar companies, related to other crises (p(SimCompan|OtherCrises) = 0.90), and environmental responsibility, related to international organizations (p(EnvironmResp|InternatOrg) = 0.79). In the last phase, media discourse was sprayed in three clusters: Environmental responsibility, similar companies and economic responsibility. Drawing on theses observations, we could answer RQ3 by arguing that associative frames were only partially related.
Agenda-setting and associative framing in news media Attention.
Overall attention. The economic dimension was the most dominant story line (70%) for the media, maybe due to the high newsworthiness that economic issues have in news media (Table 2). Attention was also focused on environmental responsibility (64%) and, as already seen, legal (50%) and philanthropic responsibility (32%) received moderate to low attention. Ethical responsibility was notoriously less newsworthy (12%), expressing the little interest news media have in moral issues. In general, topics such as negative impact (8%) or other crises (6%) were not of great interest either. As for actors, companies similar to Shell (52%) played the most relevant role, which might be related to the high attention also received by the economic dimension.
When comparing media and protest groups agendas we could say that an agenda-setting effect was not produced as not relevant points in common are found. While media reinforced the economic aspect of CSR, protest actors gave priority to environmental responsibility and to the legal aspect. Moreover, the role of governments (36%) and protest actors (28%) for media was not as relevant as it was for protest actors (55%, 80%).
Attention over time. Appendix B shows some differences in attention between phases. As already said, in 2009-2011 the predominant role of economic responsibility was slightly overshadowed (75% to 67%) by the attention gained by environmental responsibility (60%, 69%). Presumably, some events made environmental topics more newsworthy: the BP oil spill, the European Union decision on changing the carbon emission trading scheme, the UK government decision to reduce its greenhouse gas emissions. These changes in issues could also explain the increase of attention for some actors like similar companies (52%, 60%) and government (33%, 40%), as they can be seen as responsible either for the cause or the solution of such situations. Accordingly, during that period other crises also appeared for the first time in the media agenda and had their peak (19% compared to 8% in 2012-2014).
When looking at the protest groups agenda, coincidentally from 2005-2008 to 2009-2011 a decrease in the attention for economic responsibility (57%, 38%) and an increase of attention for legal responsibility (60%, 67%) occurred. Both discourses, therefore, prioritized the importance of fulfilling legal obligations over economic profits, which could also be explained by the “humanitarian gesture” that forced Shell to pay $15.5 million to the Ogoni people to settle a court case in 2009. Even though government actors gained some attention in media over time (33%, 40%, 41%), they were always less relevant than for protest actors (38%, 57%, 60%). The same occurred for armed groups who were always irrelevant for media (6%, 7%, 6%), but gained attention in protest groups agenda (3%, 19%, 16%) maybe due to the accusations against Shell for financing armed militant groups in the Niger Delta (Platform, 2010). Although reactions against Shell had a very relevant role among protest, they were slightly covered by the media (10%, 14%, 14%).
Accordingly, hypothesis 2, proposing that the increase of attention of objects in media discourses leads to an increase of awareness about those objects among protest actors, is partially accepted since important overlaps were found in 2009-2011, but no relevant similarities were identified in the other periods.
Associative framing.
Overall associations. The associative framing presented in Appendix H shows the two independent frames on which media based their overall discourse. In both the relations of environmental responsibility and economic responsibility with other issues and actors are moderate; strong associations are not found. The environmental aspect was brought into the discourse in relation to negative reputation (p(EnvironmResp|NegReput) = 0.66), and was moderately discussed when protest actors (p(EnvironmResp|ProtestActors) = 0.66) and international organizations were mentioned (p(EnvironmResp|InternatOrg) = 0.70). This could suggest that environmental issues implied negative connotations and were covered by the media in terms of protest actions and international organizations (need for) intervention.
A second cluster focuses on economic responsibility linked to reputation and CSR
(p(EconomResp|ReputLegit) = 0.66), p(EconomResp|CSR) = 0.76). The consequences of the instability of oil prices on CSR could explain this: “There is no doubt that many shortsighted businesses will stop, or even reverse, investment in sustainability and corporate social responsibility initiatives” (The Guardian, 2008).
Contrarily to protest actors, media semantic network looks simple and atomized (Appendix H and I). Protest groups also framed negative reputation and international organizations in terms of environmental responsibility (p(EnvironmResp|NegReput) = 0.88, p(EnvironmResp|InternatOrg) = 0.95), but compared to media the correlations were much stronger and were only a small part of a big network of solid connections. Contrarily to media, where economic responsibility was positioned as an independent frame poorly connected to other CSR dimensions, protest actors framed it in terms of environmental responsibility (p(EnvironmResp|EconomResp) = 0.89), which reflects the hierarchy of attention received by the two concepts in the different discourses (economic responsibility was ranked higher than environmental responsibility in media agenda).
Associations over time. Associations in media speeches had a relatively stable pattern characterized by independent frames linked to few concepts. During the first three years three concepts were defined in terms of economic responsibility: Partners
(p(EconomResp|Partners) = 0.68), reputation (p(EconomResp|ReputLegit) = 0.70), and CSR (p(EconomResp|CSR) = 0.83). “[…] Climate change mitigation became an industry backed by the world's biggest companies” (The Guardian, 2008). “The question is how can business address these challenges and still make a profit" (The Guardian, 2006). “Social responsibility plays a big role […] it is not just about costs. There can be financial rewards, too” (The Financial Times, 2005). In 2009-2011 media discourse changed into a twofold, in none of which economic responsibility was present going in line with the decline of attention it suffered during that period. On one side, the relation between international organizations with
environmental responsibility (p(EnvironmResp|IntOrg) = 0.79) was emphasized: “[the UN plans to] estimate the costs dumped on the environment by the world's 3,000 biggest public companies” (The Guardian, 2010). On the other side, other crises were mentioned in strong relation with similar companies (p(SimCompan|OtherCrises) = 0.90). During the last phase an integration of the previous three clusters occurred with small variations: Partners moved from the economic towards the environmental aspect (p(EnvironmResp|Parners) = 0.69), and reputation was less framed in terms of economic responsibility (p(EconomResp|ReputLegit) = 0.67) and more in terms of legal responsibility (p(LegalResp|ReputLegit) = 0.54), which aligns with the growing attention for the issue.
To respond to research question 2, asking for the relation between the associative frames within media and the protest actors discourses, both semantic networks were compared (see Appendix E and F for conditional probabilities). The simplicity of media speeches in comparison with protest groups has been one of the biggest differences in the overall picture. This difference disappeared as protest actors simplified their discourses, passing from a very complex network of associations in 2005-2008 to one central frame in 2009-2011 and to a twofold discourse in 2012-2014. Changes in protest actors were more drastic that in news media, but always kept environment at the core of their speeches. In 2005-2008 while economic responsibility occupied the center of media discourses, protest groups prioritized environmental responsibility over it. In the following phase both framings became closer in terms of simplicity. Both narratives might have been influenced by the BP oil spill, but while media suggested that only energy companies were related with crises and that environmental responsibility was a concern of international organizations, protest actors asked for Shell’s partners involvement. Finally, in the last period while media framed partners in terms of environmental responsibility, protest groups framed them in terms of protest actors (p(ProtestActors|Partners) = 0.91).
Stakeholder agenda-setting and associative framing in protest actors Attention.
Overall attention. Protest actors also gave a prominent place to some of the CSR dimensions over other issues. Environmental responsibility (87%) received the highest attention, followed by legal (68%) and philanthropic responsibility (64%). However, and contrarily to the corporate agenda, economic responsibility (45%) was less present than other CSR dimensions and even than negative reputation (56%). The high attention for legal responsibility might express their demand for normative accountability regarding Shell’s actions. “Shell's controversial plans to drill […] oil in the Arctic […] face a potentially damaging legal challenge […] groups has filed a legal claim to prevent Shell drilling for oil” (Greenpeace, 2010). “We need legislation to spark transition to a low-carbon economy, and smart technologies won't help much for that” (Greenpeace, 2009).
Similarly to the other agendas, the weak presence of ethical responsibility (17%) confirms that moral issues were not of interest for protest groups either. Table 2 shows how they placed themselves in the center of their discourses (80%), while for the corporation they were relatively less importance (43%). Two unexpected results attract our attention: The role of affected populations was not as dominant in protest actors discourses (47%) as it was in Shell discourses (71%), and the attention for negative impact was bigger in Shell (38%) than in protest groups speeches (33%). This last point is explained by the emphasis Shell puts to corporate measures aiming at avoiding the negative impact of its actions (“zero fatalities”)
Attention over time. Although the decrease of attention for environmental responsibility between 2005-2008 and 2009-2011 (92%, 81%), and its small recovery in 2012-2014 (87%), this issue has always been predominant in protest actors discourse
(Appendix B). Also interesting is the outstanding increase of attention for legal responsibility in the ten years (60%, 67%, 70%). The BP crisis and the court cases against Shell for its actions in Nigeria could explain not only why the attention of protest actors moved to the
legal dimension between 2005-2008 and 2009-2011, but also why governments became more important. Particularly surprisingly is the large decrease of attention for reactions against Shell observed between the first two periods (88%, 37%). An explanation could be that seeing the higher coverage media was already giving to the issue (10% to 14%), protest actors moved their attention towards different issues. During the last period we can see how reactions (37%, 40%) and environmental responsibility (81%, 87%) recovered attention.
Protest actors and Shell agendas agreed on the importance of environmental responsibility as the main issue in all the phases analyzed, but the attention for other dimensions clearly differs. The economic aspect as a fundamental dimension for the continuity of the corporation was more important for Shell than for protest groups, who considerably reduced their attention for it in the first two phases (57%, 38%). Since not big coincidences could be found between both agendas, Hypothesis 3, arguing that the increase of salience of objects in corporate discourses leads to an increase of coverage of those objects in protest actors discourses, has to be rejected.
Associative framing
Overall associations. Overall, protest groups narrative is complex for the multiple connections between concepts (Appendix I). Its main focus was on environmental
responsibility strongly related to other concepts (most correlations above 0.80). Compared to Shell semantic network, similarities in the degree of complexity are clear. However, while Shell framed negative reputation in terms of economic issues (p(EconomResp|NegReput) = 0.96), protest actors did it in terms of environmental (p(EnvironmResp|NegReput) = 0.80) or legal responsibility (p(LegalResp|NegReput) = 0.57): “[…] Not a day passes without another big business making a green pronouncement […] Climatewash - Greenwash 2.0. Simple, cheap and no real change needed - great for big business!” (Greenpeace, 2007). “So far this year, over US$ 82 million has been spent on corporate lobbying including against climate change legislation, by the oil and gas industry alone” (Greenpeace, 2009).
Associations over time. Protest actors discourse evolved from a very complex
network in 2005-2008 to simple thematic clusters in the following periods. This change is the central difference with regard to Shell narratives. While Shell maintained complex speeches, since 2009 protest actors started limiting their discourses. In 2009-2011 they focused on the environmental responsibility frame with quite a few concepts strongly related to it, and in 2012-2014 on two clusters: protest actors and environmental responsibility. Contrarily to Shell, protest actors presented the company as an actor with almost inexistent interest in the ethical dimension (p(EthicResp|Shell) = 0.12 in 2005-2008, p(EthicResp|Shell) = 0.09 in 2008-2011, 2012-2014).
An answer to research question 4 (“How are CSR associative frames in the interplay between corporation and protest actors’ discourses related with each other?) could be that Shell and protest actors narratives have a common point regarding environmental
responsibility, but present no more important correlations.
Discussion Summary of findings
To answer our main research question, “How is CSR communicatively
(de)constructed and institutionalized to build legitimacy in the dynamic interplay between corporations, media and protest actors discourses?”, and research question 1 (“How does the communicative (de)construction of CSR change over time within the identity building
process?) implying more epistemological aspects, a summary of findings is first discussed. H1 predicted that “the increase of salience of objects about CSR organizational practices in corporate discourse leads to an increase of coverage of those objects in news media discourse”. Partial agenda-building effects were identified, thus, H1 was partly accepted. On one hand, coincidences on the fluctuation of environmental and economic responsibility in Shell and media during the first two phases reflect an agenda-building effect.
These findings are in line with previous evidence of first-level agenda-building effects (Kiousis et al., 2007; Ohl et al., 1995; Parmelee, 2013) proving that corporate PR influences how media perceive some objects to be more important than others. Although media might have been influenced by Shell, in the case of economic responsibility, for example, a process of mediatization could have also occurred since the economic aspect is per se highly
newsworthy for media. As Ihlen and Pallas (2014) stated, in order to get the endorsement of media, organizations intentionally engage in meditatization processes by reinforcing similar discourses. On the other hand, the diversification of focus occurred in the last period
contradicts agenda-building theory, which can be explained by the different profiles of the newspapers: The economic focus of The Financial Times reoriented the speech towards economic responsibility once the oil spill disaster passed. As for RQ2 (“How are associative frames in the interplay between corporation and media discourses related with each other?”) besides small similarities, Shell and media narratives are not comparable in terms of
complexity in any of the periods. One explanation could be the complex “engineering” behind Shell’s communication allows the company to elaborate more complex discourses around CSR compared to media that have more diverse concepts to cover.
With regards to the narratives of media and protest actors, an agenda-setting effect was found during the period 2009-2011 (increasing attention for legal responsibility and decreasing attention for the economic aspect in both agendas), but not in the other phases. Hence, H2, stating that “the increase of salience of CSR objects in media discourses leads to an increase of awareness about those objects among protest actors”, was partly accepted. Even tough previous studies supported the agenda-setting effect (McCombs & Shaw, 1972; Carroll & McCombs, 2003; Lee & Carroll, 2011), our results showed, first, that contextual factors could attenuate it, and, second, that protest groups could be less likely to be
influenced than other publics. Regarding frames, RQ3 asked: “How are associative frames in the interplay between media and protest actors discourses related with each other?” Results
showed that both frames became similar only after 2008, as protest groups discourses were simplified. However, narratives were rather distinct in terms of associated concepts. Irrelevant associations were found between media and protest groups frames.
Finally, the study did not found a stakeholder agenda-setting effect from the protest actors to the corporation discourses, as H3 proposed (“The increase of salience of CSR objects in protest actors discourses leads to an increase of salience of those objects in the corporate discourse”). The attention for all the concepts in every phase—except for environmental responsibility—clearly differed. One big difference, for example, was the higher importance that Shell gave to the economic aspect. Since the outcomes did not support the influence of the protest groups on the corporate agenda regarding the salience of objects, H3 was rejected. RQ4 asked, “How are associative frames in the interplay between protest actors and the corporation discourses related to each other?” Overall, frames were not comparable. An interesting duality arose: The high interest of activist groups in making Shell legally accountable (legal responsibility) versus Shell’s efforts for portraying itself as a good citizen (philanthropic responsibility). This can go in line with Hunter et al. (2013) arguing that stakeholders exert a gatekeeper role provided with their own agenda and communication strategies to influence and mobilize other actors. Unlike other publics, protest actors are focused on very specific missions and goals that might be hardly modifiable.
Based on the findings, an answer to our main research question is that Shell CSR has gone through moderate different stages influenced by external discourses and events. Overall, Shell CSR has been institutionalized around specific dimensions with strong emphasis in the environmental aspect (as a basic condition for financial success), and in philanthropic responsibility. Over time, CSR has been built on very complex discourses in terms of
concepts and associations, which express the highly “engineered” CSR communication of the company oriented towards specific and solid goals, contrarily to protest actors. However, the changes support the idea that corporate issues, actors and associative framing fluctuate
according to other actors’ making sense of reality, going in line with the constitutive and communicative view of CSR (Schultz et al., 2013; Patriotta et al., 2011).
As for RQ1, at the internal level the CSR dimensions emphasized by Shell oriented the company towards the institutionalization of rules, values and practices (organizational culture). At the external front, Shell reinforced its discourses around these dimensions via institutional reports containing specific and strong claims about them (organizational image). This expresses the culture-identity-image loop (Hatch & Schultz, 2002) through which organizational identity is built. Moreover, the multilevel efforts in trying to build legitimacy at different levels could be theoretically understood as a form of institutional work: CSR provides actors a space to interplay and shape the organization (Lawrence et al., 2011).
Conclusion
The aim of the present study was to analyze the agenda and associative framing of Shell CSR during 10 years and their differences with regards to news media and protest actors’ agendas and frames. Based on an automated content and semantic networks analysis, 209 corporate pieces, 3373 news articles and 326 non-profit organizations documents were analyzed.
The study contributes to previous research by broadening the scope of CSR studies on the oil industry and extending CCO and institutional work research. By applying a semantic network analysis over high amount of documents and a longer period of time, it goes beyond the traditional understanding of agenda-setting and framing processes. The introduction of the third-level agenda-setting or “associative framing” is crucial. Accordingly, the notion of the interplay of actors in constructing reality is understood from a more complex perspective. Furthermore, the extensiveness of the study in terms of both the time period examined and the different dimensions of CSR offers important insights for deeper analysis of the evolution of public discourses as a result of social interactions. The study overcomes instrumental and political-normative perspectives of communication, and empirically supports previous
