Skip to navigation – Site map

HomeIssues6:1Emotional Expressions and Advocac...

Emotional Expressions and Advocacy Coalitions

Allegra H. Fullerton, Kayla M. Gabehart and Christopher M. Weible
p. 5-22

Abstract

While many policy process theories mention emotions, they have remained mostly unexplored theoretically and empirically, even as broader social science literature incorporates emotions into understanding policy process-related phenomena such as political beliefs and behaviors. This paper introduces the theoretical arguments and a method for studying advocacy coalitions using a combination of emotions and beliefs within the Advocacy Coalition Framework. An application is illustrated in a natural gas pipeline siting conflict in the US using data from news media coverage. The empirical results show that coalitions express emotions and beliefs differently, and that the dyadic relationship between emotions and beliefs significantly distinguishes coalitions rather than emotions by themselves. This paper takes a significant step forward in integrating emotional and belief expressions into the ACF, adding to coalition identification methods, providing a foundation for advancing theory, and contributing to the broader community of policy studies.

Top of page

Full text

Introduction

1Emotions permeate all aspects of human experience, influencing behaviors, values, and understanding. Consequently, in the study of politics and policy, there has been growing interest in studying emotions’ role in various areas such as political behavior, argumentation, the formation of political beliefs, intransigent policy conflicts, and how individuals and groups influence policy (Maia & Hauber, 2020; Pierce, 2021; Durnová, 2022; Verhoeven & Metze, 2022; Crow & Jones, 2018). We continue this trajectory by situating emotions, in conjunction with beliefs, as a viable part of researching advocacy coalitions in policy processes. This paper introduces a method of studying coalitions via emotions and beliefs and illustrates it in a conflict around the issue of siting energy infrastructure.

2To advance the study of emotions in policy process scholarship, we utilize the Advocacy Coalition Framework (ACF) (Nohrstedt et al., 2023). For about four decades, the ACF has been instrumental in enhancing our understanding of advocacy coalitions, learning, and policy change (Weible et al., 2008; Pierce et al., 2017). Applications of the ACF number in the hundreds and span the globe (Ohno, 2022; Osei-Kojo et al., 2022; Gronow et al., 2021; Ruysschaert & Hufty, 2020; Aamodt & Stendsal, 2017). It has been applied across democratic and non-democratic governance arrangements (Li & Weible, 2020; Hughes & Meckling, 2017), from subnational to international scales of government (Rietig, 2016; Elgin & Weible, 2013), and to a variety of policy issues. The ACF continues to push forward diverse methods and approaches (Elgin & Weible, 2013; Satoh et al., 2021; Howe et al., 2021) and is compatible when used in combination with other theories and approaches across social science disciplines (Ydersbond, 2018; Markard et al., 2015).

3One of the lessons from these decades of scholarship is that advances in the methods of the ACF often precede advances in its theories (Weible & Workman, 2022). We see this in the original development of legislative coding procedures (Sabatier & Jenkins-Smith, 1993; pp. 247-251), laying the foundation for using text to analyze belief systems of advocacy coalitions. Similarly, Zafonte and Sabatier (1998) brought network analysis to the ACF to reveal the relational structures of coalitions, paving the way for the study of coordination networks and belief homophily. Discourse Network Analyzer (DNA) provided a tool for coding beliefs in the public discourse (Leifeld, 2013), spurring innovative research on coalitions’ expressions in the news media and in policy-related testimonies and hearings (e.g., Kukkonen et al., 2018; Schmid et al., 2020; Nam et al., 2022). More recently, Satoh et al.’s (2021) development of the Advocacy Coalition Index provides one of many standardized procedures for comparing coalitions. The same pattern can be found outside the ACF, especially in the development of methods for analyzing institutions (Siddiki et al., 2021) and narratives (Shanahan et al., 2018). This paper continues this trend of advancing theory with sharper methods via a new starting point, through Emotion-Belief Analysis (EBA) in the public discourse.

4This paper begins by establishing the theoretical underpinnings for studying emotions and beliefs in the ACF. This includes revisiting the ACF’s model of the individual and elucidating key concepts, particularly feelings or affect and emotional expressions and their interdependence with belief systems. The methods section then furnishes the operational guidance for conducting EBA-style analyses under the ACF. As elaborated in the methods, in employing EBA, the focus is on how coalition members express their emotions and beliefs in the public discourse rather than on their actual feelings or genuine beliefs. The EBA approach draws upon similar techniques in the sociology of emotions, the study of emotions and political behavior, and recent policy-related empirical work using positivist and interpretive methods (Rogers & Robinson, 2014; Brader & Marcus, 2013; Bustinza & Witkowski, 2022; Durnová, 2019; Durnová & Weible, 2020; Verhoven & Duyvendak, 2016; Pierce, 2022; Yordy et al., 2023; Gabehart et al., 2023; Fullerton et al., 2023; Fullerton & Weible, 2024). We then illustrate how to code emotional-belief expressions in a case involving the controversial construction of a natural gas pipeline in the United States using coalition member expressions in the news media from 2014 to 2017. The conclusion lays out a future research agenda using EBA in the ACF and beyond.

The Theoretical Foundations for Studying Emotions and Beliefs in the ACF

5While the literature describes “emotions'' in myriad and inconsistent ways, we adopt specific definitions. We distinguish between affect and emotions, with affect defined as biological/physiological feelings that are often ambivalent, amorphous, and not easily observable to anyone, even those experiencing them (Mercer, 2010). We define emotional expressions (or just emotions) as affects communicated (and thus observable) through language or other behavioral phenomena (Neuman et al., 2007; Solomon, 2012; Rogers & Robinson, 2014; Durnová, 2019). The paper aims to incorporate affect and emotions into the ACF via EBA. This necessitates both theoretical elaborations for the more abstract and often unobservable concept of affect (e.g., how does the concept of affect fit within the theoretical infrastructure of the ACF?) and operational guidance for empirical research for the more concrete, observable concept of emotions (e.g., how can the phenomenon of emotions be studied under the ACF?). This section deals primarily with the theoretical elaborations and the methods section focuses on the operational guidance.

6Affect is linked to to the ACF’s theoretical assumptions about individuals. Like most policy process theories (see also Boettke & Coyne, 2005; Petridou & Mintrom, 2020; Weible, 2023), the ACF adopts a modified version of methodological individualism as its foundation for agency – that is, only people have agency, and thus, descriptions and explanations of the policy process lie with people. However, the ACF’s assumption of agency is not devoid of context. Indeed, the textbook version of the ACF assumes that its agents are subjected to the opportunities and constraints of their contexts while simultaneously attempting to influence those same opportunities and constraints to advantage their coalition and disadvantage their opponents.

7The ACF’s agents are then assumed to be boundedly rational; they are limited in their cognitive capacity to perceive, assimilate, and process information from their environment and make decisions (Simon, 1957; Jones, 2001). The critical driver of cognition – i.e., how people think – for the ACF’s boundedly rational individuals are beliefs, which are conceptualized and modeled as a three-tiered belief system featuring deep core beliefs (e.g., fundamental values and identities), policy core beliefs (e.g., general policy positions and views of problems), and secondary beliefs (e.g., instrumental beliefs for achieving policy core beliefs). The ACF describes people’s belief systems as closely entwined with people’s sense of self and as the source for forming and maintaining in-groups and out-groups called advocacy coalitions. Advocacy coalitions are thus informal alliances of government and non-government individuals bound together by their shared belief systems and against threats from opponents with different belief systems (Nohrstedt et al., 2023). Of these three belief system categories and given their pertinence to the policy issue at hand, policy core beliefs serve as the principal glue binding coalitions together (Nohrstedt et al., 2023).

8The ACF’s argument for the primacy of belief systems implicitly incorporates feelings or affect, which partly provides the mechanisms for the longevity of coalitions over time and how people in coalitions overcome the threats to collective action inherent in all political activities (see similar arguments in Sabatier & Weible, 2007, p. 197). For instance, there is a visceral connection between belief systems and the self: deep core beliefs are akin to religious convictions, and policy core beliefs consist of normative and empirical dimensions rooted in people's feelings of concern for others or even ideas or understandings pertinent to the policy subsystem (Sabatier & Jenkins-Smith, 1999, p. 133). Thus, people are intimately attached to their belief systems, which is one of the reasons why they resist change and are motivated to promote and defend them. Another instance comes from Prospect Theory (Kahneman & Tversky, 2013), which states that people remember losses to their opponents more than gains among their allies and, therefore, exaggerate the maliciousness and power of their opponents. The result is the ACF’s original "devil shift" concept, which assumes that people's fear or threats from opponents lead to exaggerating the costs imposed by their opponents' behavior and disproportionate responses (Sabatier et al., 1987). Similarly, the ACF created the “angel shift,” where people exaggerate the virtues and power of their allies (Leach & Sabatier, 2005). Thus, for the ACF, the visceral association between affect and belief systems and between affect and interpretations of allies and opponents make advocacy coalitions an intransigent political force in policy processes.

9Throughout these explanations, the ACF establishes a theoretical link between affect or feelings and individuals' belief systems. For example, individuals’ compassion for a target population or their fear of opponents’ policy proposals can motivate their political mobilization and the formation of an advocacy coalition. The primary focus is on the dual relationship between the ACF’s belief systems and affect in individuals’ minds. The interplay of belief systems and affect, including how a person thinks cognitively, builds on the sociopsychology literature (Neuman et al., 2007; Stets & Turner, 2014). For example, it echoes Mercer’s (2010) arguments about the interdependence of beliefs and affect. It also mirrors Pessoa (2008), who interconnects how people think (i.e., cognition) and affect, and the Narrative Policy Framework that stipulates that emotions precede cognition and narration (Jones et al, 2023, p. 168). While we might not know precisely how people think or feel, or whether we feel before we think or think before we feel, we know these mental processes occur almost simultaneously and essentially in tandem, making emotions central to decision-making and action (Jones, 2001; Jasper, 2011).

10Given the theoretical interdependence of affect and beliefs in the ACF, the operationalization of these concepts must follow suit. As elaborated on in the methods section, studying emotional expressions should occur in tandem with belief expressions. For example, the study of deep core, policy core, or secondary belief expressions can be identified and then linked to emotional expressions (Gabehart et al, 2023; Fullerton and Weible, 2024). For example, a statement expressing the policy core beliefs about the proposed route for a natural gas pipeline might be associated with the emotion of fear (i.e., I’m scared about the proposed pipeline route). The combination of expressed emotions and belief systems are called “emotion-belief dyads.” This paper examines three categories of policy core beliefs, one related to the pipeline project in general, one related to the risks and benefits associated with the pipeline, and one related to the pipeline’s location.

11Given the novelty of studying emotional expressions and advocacy coalitions, we offer no hypotheses about the specific frequency or patterns of emotions (e.g., anger or fear) of coalitions. Rather, we offer a case study in which we explore the operationalization of the emotion-belief dyad and provide descriptions of the results. Moreover, while coalitions might differ in some emotional expressions and policy core beliefs (e.g., an anti-pipeline coalition ought to express more anger about the pipeline compared to the pro-pipeline coalition), the coalitions might or might not differ in other emotional expressions and policy core beliefs (e.g., both coalitions might express feelings of uncertainty or obligation about the pipeline). Thus, we offer advancement in the EBA method — a reasonable goal when introducing a new method based on a single empirical case study. We then aim to build a better theory about the patterns and frequency of emotional expressions and the belief system of the ACF after several empirical applications in diverse settings and using different data sources.

Methods

Case Selection

12Our analysis uses the Nexus Gas Transmission Pipeline project in the United States as a case study of public discourse in the news media. This pipeline spans 257 miles across Ohio and Michigan and delivers natural gas to markets in the midwestern United States. News media surrounding this pipeline have been previously analyzed (You et al., 2022; 2022a; Yordy et al., 2019), which indicates that the siting process of this pipeline was one with high levels of conflict between coalitions and included several legal challenges. While the final siting decision rests with the Federal Energy Regulatory Commission (FERC), different levels of government are involved in the siting process, including local and state governments. The pipeline was constructed in 2017 and became operational in 2018.

13This pipeline conflict occurred within a federalist governance arrangement in the United States. We may see higher levels of mobilization of citizens and interest groups than in other parts of the world; however, we would likely see similarities with other Western democracies in many European countries. Much of the Nexus controversy is about the pipeline's location or siting, which are policy core beliefs in this study. Many individuals, organizations, and local governments are opposed to the pipeline running under their communities. This scenario allows coalitions to form on a somewhat one-dimensional issue around the pipeline location.

14Other patterns within this conflict have implications for our analysis. The first is that this case pits a large energy company, which primarily comprises the pro-pipeline coalition, against many different (and smaller) organizations and individuals in the anti-coalition. This diversity within the anti-coalitions manifests as different emotional-belief dyads about the pipeline.

Data Source

  • 1 News media analysis is not without its limitations, and we are mindful of these as we interpret our (...)

15The data source for identifying emotion-belief expressions used in this paper is the news media, though it could be applied to other sources of discourse. News media is a useful data source for this study because it allows for the analysis of emotional statements by allies and opponents in the pipeline controversy.1 Newspaper articles have been analyzed in ACF studies as well as other policy process studies as a means of studying policy debates and actor beliefs (Henry et al., 2022; Heikkila et al., 2019; Martin & Rose, 2003; Shanahan et al., 2008; Naziz, 2020; Schaub, 2021), and recent studies of emotions also examine news media (Verhoeven, & Metze, 2022; Yordy et al., 2023). News media are also a fitting sample for our study, as Wahl-Jorgenson (2019) suggests that news media is important for articulating how others feel, as journalists often ask sources how they “feel” as a way of building their narratives. One result is that news media is a rich source of emotional statements. Additionally, scholarship has identified news media as a viable tool for examining policy-related issues (Howland et al., 2006). Nonetheless, news media still filter information through journalists and editors, and our method below reliably identifies emotional expressions, not how actors actually feel.

16We used keyword searches of “Nexus'' and “Pipeline” in Newsbank and Nexus Uni for newspaper articles between 2014-2017 across two U.S. states. We manually reviewed all of the articles to remove duplicates or articles where the pipeline was not the subject of the article. This process resulted in 372 unique newspaper articles for our analysis. Each article was uploaded to Discourse Network Analyzer Software v2 developed by Leifeld (2019), and we manually coded each article using our emotion coding approach, which is described below. The codebook is published in Fullerton and Weible, 2024.

Textual Analysis

17The articles were then uploaded to the DNA software. From here, the authors first read the articles and then manually identified statements that conveyed a policy core belief, that is, what are the policy preferences (location) and what is the policy about (pipeline) and the seriousness of the problem (risk/benefit). The authors then identified the emotion associated with this belief as conveyed by various actors in the Nexus subsystem. Within these statements, we identified the narrating actor of the statements (who was speaking), the emotional-belief actor (who was perceived to be feeling the emotion), the emotion (what was being felt), and the belief (what the emotion was about). For example, in the statement, “I’m damn mad because people are getting hurt” coders identified anger as the emotion associated with a belief about risks and benefits. This is a self-narrated statement, and the name and affiliation of the narrating actor (and emotion-belief actor) is found elsewhere in the text. We then coded the same ten articles until we reached eighty-five percent intercoder reliability. Once the team consistently reached eighty-five percent, one author coded the remaining articles while the other checked twenty percent of their work. These two authors achieved eighty-five percent or higher intercoder reliability for coding in this analysis, meaning they found agreement in eighty-five percent of their codes.

18Narrating actors and emotional belief actors were coded using their full names as provided by the newspaper. In addition to their names, we coded their affiliation or organization they represented (if there was none, this was coded as public) and their position on the pipeline (pro, anti, neutral, undisclosed). We manually assessed the full compilation of statements to designate a person as pro, anti, or undisclosed. For example, if one actor had five statements against the pipeline, that person would be coded as a member of the anti-pipeline coalition. Alternately, if an actor had 20 statements, with 17 for and three against the pipeline, and if the total collection of statements suggested support for the pipeline, the actor would be coded as a member of the pro-pipeline coalition. If it was unclear whether the actor had a pro or anti-pipeline position, that actor would be coded as undisclosed. If an actor made an equal number of pro and anti-statements, they would be coded as neutral (e.g., four statements supporting economic growth and four against environmental damage). The classification of coalition members passed the same intercoder reliability standards as the emotion coding.

19Given that we distinguish between narrating actors and emotion-belief actors, emotional expressions can be analyzed as being narrated for the self (e.g., I feel angry) or about another (e.g., they feel angry). Analyzing self and other narrated emotions comes with distinct theoretical and analytical implications. For this analysis, we examine self-narrated statements in which the actor speaking is describing their own emotions. Overall, 55 actor categories made 380 self-narrated statements.

  • 2 The interpretation of implicit emotions requires contextual interpretation of the text. Thus, “who (...)

20Our coding approach allows for reliable coding of both explicit emotions (emotions directly stated, e.g., mad) and implicit emotions (emotions conveyed using non-emotional words, e.g., “who knows?” for uncertainty), which we accredit to the emotional thesaurus developed by Yordy et al., 20232. The emotional thesaurus was initially developed by reading news media articles and identifying all the emotion words expressed, which were then compared to each other and categorized into the 12 emotional categories (want, uncertainty, trust, approval, suffering, obligation, fear, disapproval, content, compassion, careless, and anger) The technique is similar to the method of explication that rhetoricians use and was led by an interpretive scholar. This approach allowed for the thesaurus to be built from the ground up, and as we applied the thesaurus to a new case and sometimes revised it, we found that the 12 categories continued to capture reliably and validly expressed emotions.

21Many scholars within the social sciences have developed various emotional categorical approaches, and while ours is built from the ground up, it is influenced by the work of Gordon (1990). He reminds us that as we work to classify emotions, we need to consider their microsocial structures in addition to their resemblances (Turner & Stets, 2005). Our thesaurus, and specifically how we classify explicit and implicit emotions, speaks to the necessity of considering context in our work. Because we are building the thesaurus from the emotions found in our sample, several emotion words were not included in the thesaurus at the start of the project. We continued to build the thesaurus using the following decision rule: If we found a new word that we thought might be a new explicit emotion and we wanted to add it to an emotion category, we would need to code it consistently across statements; that is, the word would have to be coded within the same category across multiple coders and statements. The team would also discuss the work and only include the word in the thesaurus if there was agreement. For example, “trauma” would always be suffering as we agreed that it expressed the same emotion of suffering across contexts. In contrast, “challenging” is absent from our thesaurus as an explicit emotion because it can be coded as suffering or disapproval, depending on the context. The authors discussed all additions to the thesaurus, and words were included in the thesaurus only when there was agreement. If the meaning of the word was context-dependent, such as “challenging,” we would still code the word, albeit implicitly. Finally, we included only the root word in our thesaurus; for instance, “agree” was taken for approval. However, “agreement,” “agrees,” and “agreed” were also coded as approval. The implicit and explicit coding allowed us to consider the context of the statement or article and provided a more accurate identification.

22For our coding, we compared our coded emotional expressions to the thesaurus. The thesaurus has 356 emotion words within the 12 categories we used for comparison. For instance, when we read the news media statement “We are concerned about the pipeline in our town,” we searched the thesaurus for “concern” and found it within the fear category. We coded the statement for both “concern” and fear and marked the emotion as explicit. For implicit statements, we coded based on the categorical emotion. For instance,” My wife and I are running out of patience,” was coded as “frustration” as the statement was under the suffering category. We coded this statement as suffering and implicit. By utilizing the thesaurus, the coders were able to reliably identify the same emotions in the statements.

Results

23Figure 1 depicts the linkages between emotional expressions and policy core beliefs via the emotion-belief dyads. The network map on the left, with the black nodes, shows the anti-pipeline coalition, and the one on the right with the white nodes shows the pro-pipeline coalition. For both network maps, the size of the nodes represents the relative proportion of emotional expressions per coalition. For example, for the anti-pipeline coalition, twenty-nine percent of the 292 self-narrated statements (or eighty-five) express the emotion of fear, more (a bigger node) than the one percent of the 292 self-narrated statements (or three) that express approval. Similarly, the size of the nodes for the policy core beliefs also represents the relative share of all statements per coalition. Tie thickness also illustrates the relative percentage of all ties between each expressed emotion and policy core beliefs per coalition. Finally, the rank order of expressed emotions and policy core beliefs is shown from largest to smallest for each coalition.

Fig. 1 Coalitions’ Emotions and Policy Core Beliefs-

Fig. 1 Coalitions’ Emotions and Policy Core Beliefs-

Source: the Authors

24For the anti-pipeline coalition, fear, dissatisfaction, and suffering represent the most common emotions as a proportion of the coalition’s total, or sixty-three percent. Most of these emotions link to the pipeline project and its risks and benefits. A small proportion relates to the pipeline location. In contrast, for the pro-pipeline coalition, approval, trust, and want represent most of its emotional expressions (sixty-four percent of the total), and its emotional expressions link to the pipeline project and its risks and benefits, with few related to the pipeline location.

25In some respects, the two coalitions diverge in the proportion of their emotional expressions. For example, fear and approval are the most and least expressed emotions for the anti-pipeline coalition; the opposite is true for the pro-pipeline coalition. In other respects, there are similarities. Both coalitions express uncertainty with similar frequency, and they rarely express other emotions, such as careless (one percent). Moreover, neither coalition expresses the emotion of content, which is not shown in Figure 2.

26Figure 2 provides an overview of the distribution of the statements and the actor categories in two pie charts for each coalition. On the left of Figure 2 is the anti-pipeline coalition comprising 292 self-narrated statements from 36 actor categories, and on the right of Figure 2 is the pro-pipeline coalition with 88 self-narrated statements from 19 actor categories.

27For the anti-pipeline coalition, the actors in the unaffiliated “public” category make up the largest slice with thirty-two percent of its 292 statements, or ninety-three. The second largest category is the Coalition to Reroute Nexus, an interest group dedicated to keeping the Nexus pipeline out of the community. Thirty contributors to the anti-pipeline coalition fall into small, unnamed slices of actors, each making seven or fewer statements or accounting for twenty-six percent (or seventy-seven) of the 292 self-narrated statements. For the pro-pipeline coalition, Nexus, the pipeline company, contributes the most statements with fifty-three percent of the total, or forty-seen. The public accounts for only seven percent of this coalition’s statements, meaning that a smaller share of vocal pro-pipeline advocates were unaffiliated.

Fig. 2 Advocacy Coalitions by Actor Category-

Fig. 2 Advocacy Coalitions by Actor Category-

Note: Percentages and node size indicate the proportion of statements per coalition for emotions and objects. Tie thinness to thickness reflects low to high number of links between emotions and policy core beliefs.

Source: the Authors

  • 3 We calculated the correlations using QAP on UCINET (Borgatti et al., 2002). To run the calculations (...)

28Lastly, we test whether the different dyad categories in Figure 1 have statistical significance. Figure 1 suggests that each coalition has a distinct footprint for its expressed emotions and associated policy core beliefs. To assess the possibility of such different footprints, we calculate the correlations and significance between coalition membership and expressed emotions and associated policy core beliefs. We use Quadratic Assignment Procedure (Krackhardt, 1992), which accounts for the lack of independence among observations and, through permutations of the existing data, calculates the likelihood that our observed patterns are due to chance or are unusual enough for us to claim a significant relationship.3

29Table 1 presents the results with a statistically significant relationship between coalition membership and expressed emotions and policy core belief dyads with a correlation coefficient of 0.30 (p<0.000). However, no significant relationship is found between coalition membership and only expressed emotions with a correlation coefficient of 0.013 (p>0.10), which underscores the importance of examining emotions and beliefs in conjunction.

Table 1. Correlating Coalitions, Emotions, and Policy Core Beliefs-

Quadratic Assignment

Procedure Correlations

Coalition Membership

Coalition Emotions and Policy Core Beliefs

0.30, p<0.000

Coalition Membership

Coalition emotions (only)

0.013, p>0.10

Correlation coefficients and statistically significance shown.

Source: the Authors

30Whereas Figure 1 splits the data and calculates the proportion of emotions expressed and the policy core beliefs per coalition, the calculations in Table 1 combine the data, which is based on the proportions of expressed emotions and policy core beliefs per coalition in Figure 1.

  • 4 To conduct the Multi-Dimensional Scaling, we first converted the matrix of fifty-five actor categor (...)

31To provide a more detailed account of how the coalitions express emotions, we examined the data using different techniques in Figure 3. Figure 3 provides a visual layout of the emotional expressions for both coalitions for the entire sample of statements using Multi-Dimensional Scaling.4 The Multi-Dimensional Scaling procedure visualizes the relationships between expressed emotions by placing them in a two-dimensional space. Emotional expressions that are close to each other are expressed by similar actor categories.

32We calculate the clusters of only the emotions (i.e., the circles in Figure 3) using Tabu Cluster Search Optimization (Glover, 1989; Glover et al., 1995). The optimal number of clusters is four at R2=0.59. Whereas the Multi-Dimensional Scale pulls and places the emotions in proximity to each other based on horizontal and vertical dimensions, Tabu Cluster Search Optimization ignores issues of dimensionality and clusters based on overall similarities in use across actor categories.

33Lastly, we include percentages from the entire sample of statements per coalition next to the nodes in the Multi-Dimensional Scaling Figure. The percentages on the left are for the anti-pipeline coalition and those on the right are for the pro-pipeline coalition. For example, the emotional expression of anger comprises 15 of the 380 statements, all (four percent) made by the anti-pipeline coalition and none (zero percent) by the pro-pipeline coalition.

34The leftmost cluster in Figure 3 confirms the emotional expressions (anger, suffering, fear, and dissatisfaction) used more frequently by the anti-pipeline coalition than by the pro-pipeline coalition. For example, the anti-pipeline coalition expresses fear in twenty-three percent of the 380 self-narrated statements, whereas the pro-pipeline coalition expresses fear in one percent of the 380 self-narrated statements. On the far right of Figure 3 lies a single cluster with the emotional expression of approval. This is the only emotion that the pro-pipeline coalition expresses more frequently than the anti-pipeline coalition. In the middle of Figure 3 lie the five emotions that both coalitions express with relatively equal frequency: obligation, want, trust, compassion, and uncertainty. The last cluster, careless at the top, is the one emotion that neither coalition expresses much. Figure 3 indicates why the correlation between coalitions and emotions is insignificant. With some emotions, the two coalitions diverge (clusters on the far right and left), and there are other emotions that are roughly the same (clusters in the middle and at the top).

Fig. 3 Emotional Expressions (n=380) as found in the Public Discourse

Fig. 3 Emotional Expressions (n=380) as found in the Public Discourse

Analysis conducted with 380 self-narrated pro- and anti-coalition statements (the entire sample). Anti percentages are on the left and pro percentages are on the right. Visual layout is presented using Multidimensional scaling with Stress = 0.12. Clustering is computed using Tabu Cluster Search Optimization with two clusters at R2= .35, three clusters at R2= .51, four clusters at R2= .59 (presented), and five clusters at R2=0.50. Percentages calculated from self-narrated emotions expressed in the sample (emotions by anti-coalition percent, emotions by pro-coalition percent).

Source: the Authors

Discussion & Conclusion

35This paper develops a better theory and method for studying affect and emotional expressions in relation to beliefs using Emotion-Belief Analysis under the ACF. In studying a siting of a natural gas pipeline, the results show that advocacy coalitions tend to express emotions in conjunction with their policy core beliefs similarly to allies and differently from opponents. The anti-pipeline coalition expressed various emotions, most often fear, dissatisfaction, anger, and suffering about the pipeline, its risks and benefits, and its location. In contrast, the pro-pipeline coalition primarily expressed approval of the pipeline and its risks and benefits. This finding suggests that the anti-coalition has both more overall statements and emotions than the pro and is perhaps indicative of wanting to expand the conflict. Further examination of EBA and coalition diversity is warranted. Surprisingly, both coalitions expressed some emotions equally, such as desire, uncertainty, compassion, trust, and obligation (though these were often attached to different policy core beliefs depending on the coalition).

36The findings show no significant relationship between advocacy coalition membership and emotional expressions without connections to policy core beliefs. Coalitions and only emotional expressions are not significant, which makes sense considering that members of two coalitions may express anger about two different things. This speaks to the theoretical and empirical importance of coding beliefs with emotions. Coalition members and their emotional expressions are only significant when contextualized via their relationship to distinct policy core beliefs.

37Overall, the empirical results show that EBA offers a more granular level of analysis to the more traditional approach to studying coalitions, which tended to focus on coalition members’ positions about a belief, which can be referred to as “position-belief dyads” (e.g., Jenkins-Smith et al., 1991; Zafonte & Sabatier, 2004; Heikkila et al., 2019). To illustrate, Jenkins-Smith et al. (1991) coded belief expressions in Congressional testimonies about the outer continental shelf oil and gas leasing policy debates. An example of their measures was whether coalition members agreed or disagreed with increased regulation. If EBA were applied to their study, an emotion-belief dyad might capture, for example, whether a coalition member expressed confidence or fear about increasing regulation. From this perspective, the emotion-belief dyads studied here offer an alternative approach to the more traditional position-belief dyad for studying advocacy coalitions.

38Given that emotion-belief dyads via EBA offer a comparable and alternative method to the more traditional position-belief dyads, one of the major contributions is an expansion of the theoretical and methodological toolkit under the ACF. In other words, EBA provides a viable tool for studying coalitions when researchers want to understand the manifestations of feelings or affect through emotional expressions in conjunction with beliefs. Of course, the potential gains in knowledge from an EBA-based approach have yet to be tapped. One of the next steps is to further develop the theory and methods of EBA’s capacity to identify, examine, and analyze belief systems, policy positions, conflicts, and more, which should provide a more detailed picture of the coalition landscape and open up new lines of research on learning and policy change. As we move forward and look to the future, the following questions seem pertinent: How do emotional expressions differ between deep core, policy core, and secondary beliefs? How do emotional expressions correspond to coalition formation, coordination patterns, stability, and other behaviors? How do emotional expressions differ between nascent and mature subsystems? How do emotional expressions affect policy learning and policy change?

39While not attempted in this paper, the EBA might apply to other policy process theories, such as the Narrative Policy Framework (Jones et al., 2023) or the Social Construction Framework (Schneider & Ingram, 1993). For example, linking different emotional expressions, such as fear, to people or actors might be another way to identify the Narrative Policy Framework’s villain character. We also anticipate that EBA, perhaps with further modifications, could be conducive to conducting an amalgamation of interpretive and positivist-style analyses of emotions in policy processes, as advocated by Durnová and Weible (2020) and partially illustrated in Yordy et al. (2023). There is also potential in linking EBA to the interpretive study of emotions and policy conflicts, as shown by Verhoeven & Duyvendak, Durnová, 2022, and Verhoeven & Metze (2022).

40Incorporating emotions in ACF and policy process research is an essential step in the continued evolution of policy process theories, especially to better reflect human experience. This paper joins others in arguing that emotions and beliefs are phenomena that researchers can tackle with systematic and reliable methods (Jasper 2011; Rogers & Robinson, 2014; Kim et al., 2021; Fullerton and Weible, 2024). The stakes could not be higher, as feelings and emotional expressions are essential to understanding political behavior, conflict, and decision-making in an increasingly politically polarized world. For example, research has shown that individuals who vote for far-right candidates feel that they compete with people of color, immigrants, and other groups, and that they are excluded democratically, and thus respond to xenophobic rhetoric (Hochschild, 2018, Gabehart, 2022; Shucksmith et al., 1994). These phenomena, and those related to emotions are critical to integrate into coalition studies, and EBA is a tool that opens up this line of inquiry. Whether or not this is true is not the point—feeling influences beliefs and behavior and vice versa. When our theories are unable to grasp these feelings and expressed emotions or overlook them altogether, we are prevented from beginning to understand the political crises facing humanity today.

Top of page

Bibliography

Aamodt, S., & Stensdal, I. (2017). Seizing policy windows: Policy influence of climate advocacy coalitions in Brazil, China, and India, 2000-2015. Global Environmental Change; 46, 114-125.

Ahmed, S. (2004). Collective feelings: Or, the impressions left by others. Theory, Culture & Society, 21(2), 25-42.

Boettke, P. J., & Coyne, C. J. (2005). Methodological individualism, spontaneous order and the research program of the Workshop in Political Theory and Policy Analysis. Journal of Economic Behavior & Organization, 57(2), 145-158.

Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies.

Brader, T., & Marcus, G. E. (2013). Emotion and Political Psychology. In Oxford Handbook of Political Psychology. Oxford University Press.

Bustinza, M. A., & Witkowski, K. (2022). Immigrants, deviants, and drug users: A rhetorical analysis of President Trump’s fear-driven tweets during the 2019 government shutdown. Policy & Internet.

Crow, D., & Jones, M. (2018). Narratives as tools for influencing policy change. Policy & Politics, 46(2), 217-234.

Cappella J. N., & Jamieson, K. H. (1997). Policy makes mass politics. Annual Review of Political Science, 15, 333-351.

Dunaway, J. (2011). Institutional influences on the quality of campaign news coverage. Journalism Studies, 12(1), 27-44.

Durnová, A. (2019). Understanding emotions in post-factual politics: Negotiating truth. Edward Elgar Publishing.

Durnová, A. P., & Weible, C. M. (2020). Tempest in a teapot? Toward new collaborations between mainstream policy process studies and interpretive policy studies. Policy Sciences, 53(3), 571-588.

Durnová, A. (2022). Making interpretive policy analysis critical and societally relevant: emotions, ethnography, and language. Policy & Politics50(1), 43-58.

Elgin, D. J., & Weible, C. M. (2013). A Stakeholder Analysis of Colorado Climate and Energy Issues Using Policy Analytical Capacity and the Advocacy Coalition Framework. Review of Policy Research, 30(1), 114-133.

Fullerton, A. H., Gabehart, K. M., Yordy, J., & Weible, C. M. (2023). Analysing emotional discourse among allies and opponents in the news media. Emotions and Society, 1-19.

Fullerton, A. H., & Weible, C. M. (2024). Examining emotional belief expressions of advocacy coalitions in Arkansas’ gender identity politics. Policy Studies Journal, 52(2), 369-389.

Gabehart, K. M. (2022). Rural society, democratic exclusion, and the cultural divide: Moving towards a research agenda of ruralness. Public Integrity.

Gabehart, K. M., Fullerton, A. H., Crawford, A. M., & Weible, C. M. (2023). How are emotions and beliefs expressed in legislative testimonies? An advocacy coalition approach. Review of Policy Research.

Gordon, S. L. (1990). Social structural effects on emotions. In T. D. Kemper (Ed.), Research Agendas in the Sociology of Emotions (pp. 145-179). Albany State University of New York Press.

Gronow, A., Brockhaus, M., Di Gregorio, M., Karimo, A., & Ylä-Anttila, T. (2021). Policy learning as complex contagion: how social networks shape organizational beliefs in forest-based climate change mitigation. Policy Sciences54(3), 529-556.

Gronow A., Ylä-Anttila T., Satoh K., & Weible C. M. (2022). Of devils, angels and brokers: how social network positions affect misperceptions of political influence. J Eur Public Policy, 12, 1-24.

Heikkila T., Berardo R., Weible C. M., & Yi H. (2019) A comparative view of advocacy coalitions: Exploring shale development politics in the United States, Argentina, and China. Journal of Comparative Policy Analysis: Research and Practice21(2), 151-166.

Henry, A. D., Ingold, K., Nohrstedt D., & Weible, C. M. (2022). Advocacy coalition framework: advice on applications and methods. Methods of the policy process, 105-136.

Hochschild, A. R. (2018). Strangers in Their Own Land. The New Press.

Howe, A. C., Tindall D. B., & Stoddart, M. C. J. (2021). Drivers of tie formation in the Canadian climate change policy network: belief homophily and social structural processes.  Social Networks, 75, 107-117.

Howland, D., Becker, M. L., & Prelli, L. J. (2006). Merging Content Analysis and the Policy Sciences: A System to Discern Policy-Specific Trends from News Media Reports. Policy Sciences, 39, 205-231.

Hughes, L., & Meckling, J. (2017). The politics of renewable energy trade: The US-China solar disputeEnergy Policy, 105, 256-262.

Jasper, J.M. (2011). Emotions and social movements: Twenty years of theory and research. Annual Review of Sociology 37(1), 285-303.

Jones, B. D. (2001). Politics and the architecture of choice: Bounded rationality and governance. University of Chicago Press.

Jones, M. D., Smith-Walter, A., McBeth, M. K., & Shanahan, E. A. (2023). The Narrative Policy Framework. In C. M. Weible (Ed.), Theories of the policy process (5th ed.). Routledge.

Kim, S. Y., Ganesan, K., Dickens, P., & Panda, S. (2021). Public Sentiment toward Solar Energy—Opinion Mining of Twitter Using a Transformer-Based Language Model. Sustainability, 13(5), 2673.

Krackhardt, D. (1992) The Strength of Strong Ties: The Importance of Philos in Organizations. Journal of Networks and Organizations: Structure, Form, and Action, 216, 239.

Kukkonen, A., Ylä-Anttila, T., Swarnakar, P., Broadbent, J., Lahsen, M., & Stoddart, M. C. J. (2018). International organizations, advocacy coalitions, and domestication of global norms: Debates on climate change in Canada, the US, Brazil, and India. Environmental Science & Policy, 81, 54-62.

Lawrence R. G. (2000). Game-Framing the issues: Tracking the strategy frame in public policy news. Political Communication, 17(2), 93-114.

Leifeld, P. (2013). Reconceptualizing major policy change in the advocacy coalition framework: A discourse network analysis of German pension politics. Policy Studies Journal, 41(1), 169-198.

Leifeld, P. (2019). Discourse Network Analyzer. https://github.com/leifeld/dna/releases

Li, W., & Weible, C. M. (2021). China’s Policy Processes and the Advocacy Coalition Framework. Policy Studies Journal, 49(3), 703-730.

Maia, R., & Hauber, G. (2020). The emotional dimensions of reason-giving in deliberative forums. Policy Sciences53(1), 33-59.

Markard, J., Suter, M., & Ingold, K. (2016). Socio-technical transitions and policy change – Advocacy coalitions in Swiss energy policy. Environmental Innovation and Societal Transitions, 18, 215-237.

Martin, J. R., & Rose, D. (2003) Working with discourse. Continuum.

Mercer, J. (2010). Emotional Beliefs. International Organization, 64(1), 1-31.

Nam, A., Weible, C. M., & Park, K. (2022). Polarization and frames of advocacy coalitions in South Korea’s nuclear energy policy. Review of Policy Research, 39(4), 387-410.

Naziz, A. (2020). Sustainable Development Goals and Media Framing: An Analysis of Road Safety Governance in Bangladeshi Newspapers. Policy Sciences, 53, 759-777.

Neuman, W. R., Marcus, G. E., MacKuen, M., & Crigler, A. N. (Eds.). (2007). The affect effect: Dynamics of emotion in political thinking and behavior. University of Chicago Press.

Nohrstedt, D., Ingold, K., Weible, C. M., Koebele, E. A., Olofsson, K. L., Satoh, K., & Jenkins-Smith, H. C. (2023). The Advocacy Coalition Framework. Progress and Emerging Areas. In C. M. Weible (Ed.), Theories of the policy process (5th ed.). Routledge.

Ohno, T. (2022). Advocacy Coalition Framework in Environmental Governance Studies: Explaining Major Policy Change for A Large Dam Removal in Japan. International Review of Public Policy, 4(1).

Osei-Kojo, A., Ingold, K., & Weible, C. M. (2022). The Advocacy Coalition Framework: Lessons from applications in African countries. German Political Science Quarterly, 63, 181-201.

Pessoa, L. (2008). On the relationship between emotion and cognition. Nature Reviews Neuroscience, 9(2), 148-158.

Petridou, E., & Mintrom, M. (2021). A Research Agenda for the Study of Policy Entrepreneurs. Policy Studies Journal, 49(4), 943-967.

Pierce, J. J. (2021). Emotions and the policy process: Enthusiasm, anger and fear. Policy & Politics, 49(4), 595-614.

Pierce, J. J., Miller‐Stevens, K., Hicks, I., Castaneda Zilly, D., Rangaraj, S., & Rao, E. (2022). How anger and fear influence policy narratives: Advocacy and regulation of oil and gas drilling in Colorado. Review of Policy Research, Early View.

Powers, A., & Fico, F. (1994). Influences on use of sources at large U.S. newspapers. Newspaper Research Journal, 15(4).

Rietig, K. (2016). The power of strategy: Environmental NGO influence in international climate negotiations. Global Governance, 22(2), 269-288.

Rogers, K. B., & Robinson, D. T. (2014). Measuring affect and emotions. In J. E. Stets & J. H. Turner (Eds.), Handbook of the sociology of emotions, Vol. 2, pp. 283–303). Springer Science + Business Media.

Ruysschaert, D., & Hufty, M. (2020). Building an effective coalition to improve forest policy: Lessons from the coastal Tripa peat swamp rainforest, Sumatra, Indonesia. Land Use Policy, 99, 103359.

Sabatier, P., & Jenkins-Smith, H. (1999) The Advocacy Coalition Framework. In P. Sabatier (Eds.), Theories of the policy process (pp. 117-166). Westview Press

Satoh, K., Gronow, A., & Ylä-Anttila, T. (2021). The Advocacy Coalition Index: A new approach for identifying advocacy coalitions. Policy Studies Journal, 51(1),187-207.

Schaub, S. (2021). Public contestation over agricultural pollution: A discourse network analysis on narrative strategies in the policy process. Policy Sciences, 54,783-821.

Schmid, N., Sewerin, S., & Schmidt, T. S. (2020). Explaining Advocacy Coalition Change with Policy Feedback. Policy Studies Journal, 48(4), 1109-1134.

Schneider, A., & Ingram, H. (1993). Social Construction of Target Populations: Implications for Politics and Policy. The American Political Science Review, 87(2), 334-347.

Shanahan, E.A., McBeth, M.K., Hathaway, P.L., and Arnell, R.J. (2008). Conduit or contributor? The role of media in policy change theory. Policy Sciences, 41,115.

Shanahan, E. A., Jones, M. D., & McBeth, M. K. (2018). How to conduct a Narrative Policy Framework study. The Social Science Journal55(3), 332-345.

Shucksmith, M., Chapman, P., Clark, G., & Black, S. (1994). Social welfare in rural Europe. Journal of Rural Studies, 10(4), 343-356.

Simon, H. A. (1957). Models of man; social and rational. Wiley.

Solomon, T. (2012). I wasn’t angry, because I couldn’t believe it was happening’: Affect and discourse in responses to 9/11. Review of International Studies, 38 (4), 90728.

Soroka, S. (2012). The gatekeeping function: Distributions of information in media and the real world. Journal of Politics, 74(2),514-528

Soroka, S., & Wlezien, C. (2022) Information and Democracy Public Policy in the News. Cambridge University Press.

Stets, J. E., & Turner, J. H. (Eds.). (2014). Handbook of the Sociology of Emotions: Volume II. Springer Netherlands.

Timmermans, S., & Tavory, I. (2022). Data analysis in qualitative research: Theorizing with abductive analysis. University of Chicago Press.

Turner, J. H., & Stets, J. E. (2005). The sociology of emotions. Cambridge University Press.

Verhoeven, I., & Metze, T. (2022). Heated policy: Policy actors’ emotional storylines and conflict escalation. Policy Sciences, 55(2), 223-237.

Verhoeven, I., & Duyvendak, J. W. (2016). Enter emotions. Appealing to anxiety and anger in a process of municipal amalgamation. Critical Policy Studies, 10(4), 468-485.

Wahl-Jorgensen, K. (2019) Emotions, Media, and Politics. Cambridge.

Weible, C. M., Sabatier, P. A., & McQueen, K. (2009). Themes and Variations: Taking Stock of the Advocacy Coalition Framework. Policy Studies Journal, 37(1), 121-140.

Ydersbond, I. M. (2018). Power through collaboration: Stakeholder influence in EU climate and energy negotiations. International Negotiation, 23(3), 478-514.

Yordy, J., Durnová, A., Weible, C. M. (2023). Fear in Context: Exploring emotional discourses about COVID-19 Protests. Administrative Theory & Praxis.

Yordy, J., You, J., Park, K., Weible, C. M., & Heikkila, T. (2019) Framing contests and policy conflicts over gas pipelines. Review of Policy Research, 36(6), 736-756.

You, J., Heikkila, T., Weible, C. M. Kim, S., Park, K., Yordy, J., & Smolinski, S. L. (2022). The distribution of conflict and attention across energy infrastructure. Public Administration, 1-22.

You, J., Weible, C. M., & Heikkila, T. (2022a). Exploring instigator and defender policy scenarios in the siting of energy infrastructure. Politics & Policy, 50(1), 8-32.

Zafonte, M., & Sabatier, P. (2004). Short‐Term Versus Long‐Term Coalitions in the Policy Process: Automotive Pollution Control, 1963–1989. Policy Studies Journal, 32(1), 75-107.

Zoch, L. M., & Turk, J. V. (1998). Women Making News: Gender as a Variable in Source Selection and Use. Journalism & Mass Communication Quarterly, 75(4), 762-775.

Top of page

Appendix

Thesaurus categories from (Fullerton et al, 2023; Gabehart et al., 2023; Yordy et al., 2023)

Top of page

Notes

1 News media analysis is not without its limitations, and we are mindful of these as we interpret our results. Newspapers can have selection bias for coverage (Powers & Fico, 1994; Zoch & Turk, 1998; Lawrence 2000; Dunaway 2011) and can be sensationalized (Cappella & Jamieson, 1997; Soroka, 2012, Soroka & Wlezien, 2022). Newspapers tend to prioritize reporting on conflict (indicative of how we see more coverage of the anti-side in our results) and coalition members with media training can dominate the news cycle. Despite these limitations, news media offer rich sources of emotional data due to their tendencies to use emotions in their narratives (Wahl-Jorgensen, 2019).

2 The interpretation of implicit emotions requires contextual interpretation of the text. Thus, “who knows?” might not always signify the same emotional category.

3 We calculated the correlations using QAP on UCINET (Borgatti et al., 2002). To run the calculations, we organized the data into the 55 actor categories and the coalition membership (a 55x1 matrix) and the 55 actor categories and the expressed coalition emotions and policy core beliefs (a 55x13 matrix), and the expressed coalition emotions (a 55x11 matrix). Using exact matches, we converted the coalition membership vector to a matrix (55x55). Using Pearson's product-moment correlations, we converted the coalitions' emotional expressions, policy core beliefs, and just emotions to square matrixes (55x55).

4 To conduct the Multi-Dimensional Scaling, we first converted the matrix of fifty-five actor categories by eleven emotions into an 11x11 matrix using Pearson’s product-moment correlations. The Multi-Dimensional Scaling procedure was then run on UCINET (Borgatti et al., 2002). We also use UCINET and the same 11x11 matrix to calculate the Tabu Cluster Search Optimizations.

Top of page

List of illustrations

Title Fig. 1 Coalitions’ Emotions and Policy Core Beliefs-
Credits Source: the Authors
URL http://journals.openedition.org/irpp/docannexe/image/4120/img-1.png
File image/png, 59k
Title Fig. 2 Advocacy Coalitions by Actor Category-
Caption Note: Percentages and node size indicate the proportion of statements per coalition for emotions and objects. Tie thinness to thickness reflects low to high number of links between emotions and policy core beliefs.
Credits Source: the Authors
URL http://journals.openedition.org/irpp/docannexe/image/4120/img-2.png
File image/png, 59k
Title Fig. 3 Emotional Expressions (n=380) as found in the Public Discourse
Caption Analysis conducted with 380 self-narrated pro- and anti-coalition statements (the entire sample). Anti percentages are on the left and pro percentages are on the right. Visual layout is presented using Multidimensional scaling with Stress = 0.12. Clustering is computed using Tabu Cluster Search Optimization with two clusters at R2= .35, three clusters at R2= .51, four clusters at R2= .59 (presented), and five clusters at R2=0.50. Percentages calculated from self-narrated emotions expressed in the sample (emotions by anti-coalition percent, emotions by pro-coalition percent).
URL http://journals.openedition.org/irpp/docannexe/image/4120/img-3.png
File image/png, 20k
URL http://journals.openedition.org/irpp/docannexe/image/4120/img-4.png
File image/png, 125k
Top of page

References

Bibliographical reference

Allegra H. Fullerton, Kayla M. Gabehart and Christopher M. Weible, Emotional Expressions and Advocacy Coalitions International Review of Public Policy, 6:1 | 2024, 5-22.

Electronic reference

Allegra H. Fullerton, Kayla M. Gabehart and Christopher M. Weible, Emotional Expressions and Advocacy Coalitions International Review of Public Policy [Online], 6:1 | 2024, Online since 15 April 2024, connection on 17 July 2025. URL: http://journals.openedition.org/irpp/4120; DOI: https://doi.org/10.4000/11whq

Top of page

About the authors

Allegra H. Fullerton

University of Colorado Denver School of Public Affairs, Center for Policy and Democracy, USA.
Allegra.fullerton@ucdenver.edu
https://orcid.org/0000-0003-2783-0824

By this author

Kayla M. Gabehart

Dept. of Social Sciences, Michigan Technological University, USA.
Kgabehar@mtu.edu
https://orcid.org/0000-0002-5719-231X

Christopher M. Weible

University of Colorado Denver School of Public Affairs, Center for Policy and Democracy
Chris.weible@ucdenver.edu
https://orcid.org/0000-0003-2241-5891

By this author

Top of page

Copyright

CC-BY-4.0

The text only may be used under licence CC BY 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search