Introduction

Public consultations are an integral part of policymaking. Organizations like the European Commission or governments routinely ask for the input of citizens and stakeholders.

The dominant perspective on public consultations is resource exchange. In exchange for access to the policymaking process, stakeholders contribute political and technical information in public consultations. Policymakers, in turn, hope to gain expertise and information about political support, legitimize policies and cultivate a reputation of responsiveness (Binderkrantz et al., 2022; Braun & Busuioc, 2020). Thus, public consultations can be part of the toolbox of collaborative governance (Ansell & Gash, 2008).

Most empirical studies focus either on the types of stakeholders participating (Beyers & Arras, 2019; Rasmussen & Carroll, 2014) or on the information contained and the framing used in contributions (Eising et al., 2015).

Our article argues that consultation contributions can additionally be analyzed with regard to their emotional content. Empirical accounts stress that contributions to public consultations are emotional (Crompton, 2015). Political psychology highlights that different emotions are related to different behavioral dispositions, which are important for the mobilization of political action (Maia & Hauber, 2020; Pierce, 2021). If we elucidate which emotions a public consultation elicits, we get a more comprehensive picture about the political conflict a policy generates and the antagonism between different coalitions (Cairney & Weible, 2017; Vogeler & Bandelow, 2018).

Thus, the research question of this article is how many and which emotions are displayed in contributions to public consultations.

We formulate two tentative expectations. The first expectation is that citizens display more emotional terms in their contributions than corporate actors. The second expectation holds that consultation contributions display more emotions if they target a concrete policy proposal, rather than the overall policy framework.

We use German electricity grid construction as a case. Germany has a tradition of interest intermediation in corporatist networks (Leifeld & Schneider, 2012) and parliamentary hearings (Eising & Spohr, 2017), but public consultations are a new phenomenon. However, to accelerate the energy transition, the German legislator has instituted a consultation of electricity grid planning. Organized interests and citizens contribute their opinion on the energy transition and on planned power lines (Fink & Ruffing, 2019). Thus, we have a case in which both citizens and corporate actors contribute, a whole range of emotions may be present, and that concerns both abstract policy goals and concrete policies.

To assess the emotional content of consultation contributions, we use a sentiment dictionary (Klinger et al., 2016) indicating seven emotions: anger, disgust, fear, joy, sadness, surprise and contempt. This is not to say that there are only seven emotions—the underlying dimensions and the relations between prototypical emotions are contested (Cowen & Keltner, 2017; Marcus et al., 2006; Rozin et al., 1999). However, these seven emotions form a useful categorization of emotions that might be expressed in our policy field under study. The categorization allows us to elucidate which emotions are prevalent. The types of emotions can then be related to different forms of political action.

Both expectations can be corroborated. The contributions of citizens contain more emotional terms, especially expressing fear. Moreover, there may be a relationship between the referral to a concrete power line and an increase in emotional terms. This is due to a change of the emotional profile of contributions: If a power line is mentioned, contributions contain less joy and surprise, but more fear and sadness.

This result contributes to several discussions. First, for the case of Germany, our analysis shows that power line construction is an emotional issue, with fear and sadness as the main emotions, but not dominated by “angry citizens” (Kostka & Anzinger, 2016). We also corroborate the argument that a positive view of the energy transition is combined with negative emotions toward specific power lines (Mueller et al., 2019a). Second, for the literature on public consultations, we show that contributions need not only be analyzed with regard to their policy positions. Current applications use hand coding (Eising & Spohr, 2017) or scaling algorithms (Klüver, 2012) to determine stakeholder positions. But the question whether consultations generate legitimacy and contribute to a positive reputation of the consulting organization (Braun & Busuioc, 2020) can also be addressed with regard to the emotions that surface in the contributions. Third, we show a way to measure the emotional reactions that a policy generates. Many policy theories use the intensity of conflict to predict outcomes (Weible, 2005, p. 464) or argue that collaborative governance fails if it “builds on a history of rancor” (Ansell & Gash, 2008, p. 553). Our contribution shows a way to assess the pattern of antagonism in a policy field, going beyond the beliefs and positions of the actors, and instead assessing the pattern of emotions a policy generates.

The article is structured as follows: The second section summarizes the literature on public consultations. We then generate expectations about the emotional content of consultation contributions. Section three discusses the research design. Section four comprises the empirical analysis. Section five concludes.

Theory: Consultations as venue of emotional expression?

Public consultations are a standard tool of policymaking. In particular, agencies that rely on a technical reputation and output legitimacy increasingly conduct public consultations (Braun & Busuioc, 2020). The reasons are manifold: Consultations institutionalize exchange between policymakers and the public. Policymakers do not have enough information to design policies; instead, organized interests and the broader public offer this information and demand access to the policy process in exchange. Thus, consultations are institutionalized arenas for the exchange of information about technical issues, but also about political support (Bouwen, 2002; Fraussen et al., 2020). Agencies furthermore hope to mobilize legitimacy for their policies, and gain a reputation for adhering to moral–legal norms (Arras & Braun, 2018). In a broader framework, consultations are thought to contribute to collaborative governance (Ansell & Gash, 2008).

Stakeholders participating in consultations aim to influence policies, and draw attention to their issues (Binderkrantz, et al., 2022). A large strand of scholarship researches who participates in public consultations (Rasmussen & Carroll, 2014), and researches the variety of their strategies (Klüver, 2012).

This scholarship mostly analyzes information about policy positions that stakeholders contribute to public consultations. Similarly, a major element of the Advocacy Coalition Framework (ACF) is the intensity of conflict, expressed by the extremity of beliefs of competing coalitions (Jenkins-Smith et al., 1991; Weible, 2005).

However, public consultations take place in emotional situations. For example, in environmental politics, “[a]ttitudes about the environment are expressed by people in economic, ethical and esthetic terms that are laden with deep emotions” (Randolph & Bauer, 1999, pp. 187, italics our own). Similarly, Wälti et al. argue that actors in Swiss drug policy avoid incorporating drug users in their policy networks, as they are seen as too emotional (Wälti et al., 2004, pp. 95, 102).

For political psychology, the attempt to distinguish between “information” and “emotions” is a futile endeavor. There is no “pure” rationality; instead, emotions guide information-seeking behavior and judgment.

One of the main debates in psychology is on the dimensionality of emotions. One the one hand, there is evidence that all emotions can be traced back to underlying dimensions such as valence and arousal (Cowen & Keltner, 2017; Diener & Iran-Nejad, 1986; Marcus, 2002; Russell, 2003) and that emotions of similar valence tend to co-occur.

On the other hand, even if we assume that all emotions are related on underlying dimensions, it has proved useful for political psychology to distinguish prototypical emotions, based on different appraisals of a situation (Frijda, 1987; Ortony et al., 1988; Smith & Ellsworth, 1985). Thus, emotions are to some extent socially constructed (Pierce, 2021), and it is important how actors interpret them. For example, the “moral emotions” contempt, anger and disgust that have the same negative valence are seen as reactions to violations of different aspects of the moral order (Haidt, 2003; Rozin, et al., 1999).

Most research has been on the differential effects of fear and anger—two emotions sharing the same negative valance, but differing in their appraisal of the situation (Lerner & Keltner, 2000; Ortony, et al., 1988). A robust result is that fear enhances information-seeking behavior (Brader & Marcus, 2013, p. 178; MacKuen et al., 2010), but impedes political mobilization (Arnold, 2021). Anger, on the other hand, is a predictor of political mobilization (Valentino et al., 2011; Wagner, 2014), but depresses information-seeking (Brader & Marcus, 2013, p. 185).

For public policy research, the focus on emotions translates into the postulate that emotions offer a heuristic to decide under bounded rationality (Cairney & Weible, 2017; Jones & Baumgartner, 2005, p. 16; Pierce, 2021). Mirroring the debate about the structure of emotions in psychology, some conceptualizations look at one dimension of emotions. The idea of the valence of policies (Cox & Béland, 2013) implies that some policy ideas have a more positive valence, and are therefore more readily accepted. Other conceptualizations use prototypical emotions like anger and fear, and link them to policy outcomes (Maor, 2016, 2017).

This literature has up to now not looked into consultations as a place where the emotional quality of policies becomes visible. Our article is an attempt to conceptualize the emotional content of consultation contributions, and offer theoretical speculations on its variation.

Our first expectation builds on the observation that consultations are used by individuals as well as by corporate actors. Obviously, even in corporate actors, individuals write the contributions. However, interest organizations may be accustomed to participating in consultations (Binderkrantz et al., 2021, pp. 475–476; Klüver, 2012). Their staff is trained to use a dispassionate style. In larger organizations, texts by several actors are compiled to form an overall contribution. For many domains, specific wordings (e.g., to express an objection) are standardized. These procedures are embedded in a culture that depicts organizations as emotion-free (Meyer et al., 1997). All these steps presumably cause consultation contributions by organizations to be “whitewashed” from emotional content.

Citizens, on the other hand, have less resources and are not embedded in an organizational structure. Moreover, citizens may feel powerless and frustrated if they have the impression that they have no meaningful impact. Many empirical accounts stress the emotional involvement of citizens (Cheyne & Comrie, 2002; Jewell & Bero, 2007; Randolph & Bauer, 1999; Zilles & Marg, 2022), and studies in the advocacy coalition tradition show that nongovernmental actors hold more extreme beliefs about policies (Jenkins-Smith, et al., 1991; Vogeler & Bandelow, 2018). Our argument also draws on the claim that actors that lack resources rely on attention-seeking strategies and moral reasoning to gain influence (Beyers, 2004).

Thus, we argue that consultation contributions by citizens contain more emotions than contributions by professional actors. This does not mean that citizens are more emotional than corporate actors. However, individuals working for corporate actors are bound by organizational norms to avoid the expression of emotions in written contributions, and citizens might have more incentives to use emotional language.

Expectation 1 Consultation contributions by citizens contain more expressions of emotions than contributions by corporate actors.

Our second expectation is that expressions of emotions about policies are strongest when policies are concrete and their implications become visible. This result is established in research on the deployment of renewable energy (Devine-Wright & Howes, 2010; Mueller, et al., 2019a). The argument is that people have an emotional “place attachment” to their surroundings (e.g., perceiving the environment as “scenic beauty” or a “run-down industrial town” (Devine-Wright & Howes, 2010)). Hence, not all disruptions to place attachment are seen as a threat (a “run-down industrial town” may profit from new infrastructure). But only the concrete policy (e.g., the power plant) generates the event that threatens the emotional bonds to the place (Devine-Wright & Howes, 2010, p. 271). Hence, the discrepancy between overall acceptance of renewable energy and local protests.

For the question of infrastructure-related policies, the differentiation between abstract framework and concrete policies is easy: Once policies comes down to the local level—their repercussions become visible—actors can identify a threat—or an opportunity—that can be connected to an emotion.

However, we think that the idea is generalizable. Our argument is that emotions become stronger once actors determine the costs and benefits a policy generates for them. This may be correlated with the stage of the policy cycle—the closer to implementation, the easier it becomes to determine winners and losers. But it is more fruitful to think of levels of abstraction that can be present at any stage of the policy process. Any policy contains paradigms—e.g., promote renewable energy—that may have a positive appeal (Cox & Béland, 2013). Yet, on the basis of this paradigm, actors cannot determine how they are affected. The policy furthermore contains specific instruments—e.g., using regulatory or redistributive instruments to promote renewables—and actors now have better notions how they might be affected. But only once the specific settings of these instruments is known—e.g., precise rules for subsidies—can actors assess how a policy affects them.

Hence, our second expectation is that emotional responses will be stronger when consultation contributions target concrete policies. What kind of emotions these are is an open empirical question that depends on the cognitive appraisal of the policy, e.g., whether it is seen as beneficial, threatening or violating fairness norms.

Expectation 2 Consultation contributions that target concrete policies contain more expressions of emotions than contributions that target the overall policy framework.

Case, research design and methods

We use the German consultation of electricity power lines as a test case to assess whether our expectations provide an accurate description of reality. The German legislator redesigned the institutions of electricity grid planning in 2011. Before 2011, the planning had been conducted in corporatist networks between the government and the utilities. The new procedure obliges the private German Transmission Systems Operators (TSOs) and the Federal Network Agency (FNA) to conduct public consultations when planning grid projects. The planning regime is part of the German energy transition. In order to integrate renewable energies, new high-voltage lines are needed. Previous experience had shown that local resistance is considerable (Fink & Ruffing, 2019; Mueller, et al., 2019a); thus, the legislator sought a method of accelerating grid planning and increasing public acceptance.

The public consultation is similar to the online consultations of the European Commission or the US notice and comment procedure: The TSOs and the FNA publish a list of planned power lines and their rationales online, and the public can respond online or via mail (Fink & Ruffing, 2019). For Germany, these public consultations are an institutional innovation. Legal scholarship knows the “Öffentlichkeitsbeteiligung” for the public concretely affected by a project, and German parliamentary committees have institutionalized hearings, but the usual route of interest intermediation in Germany is through closed policy networks (Leifeld & Schneider, 2012).

Our case offers several features that make it ideal to assess our expectations. First, the consultations are low-threshold and open to a variety of stakeholders. Second, the consultations are formulated broadly and concern the overall policy framework—the energy transition—as well as concrete policies—proposed power lines. Third, the consultation has been held in 2012, 2013, 2014, 2015 and 2017 by the TSOs and the FNA, and we have ample data on contributions.

The dataset contains all contributions to the consultations of the TSOs and the FNA from 2012 to 2017, in total 36,556 full texts. The data were scraped from the consultation website (for the TSOs) or provided by the FNA. Based on text similarity measures, we identified form letters, drafted by citizen’s initiatives and signed by individual citizens. As we are interested in original contributions, written—and not just signed—by the contributing actor, we removed these form letters from the dataset. Footnote 1 The final dataset contains 7,335 distinct contributions.

The main variables to explain the variation of emotional content are the type of actor contributing (expectation 1) and the question whether a concrete power line is mentioned (expectation 2). We used a dictionary to code which type of actor submitted a contribution. Based on a list of the organization names, we differentiated between citizen, citizens’ initiative, company, industry association, local government, Länder government, federal government, science, parties, districts and others. Footnote 2 Similarly, we used dictionary coding to code whether a power line is mentioned in the contribution. This procedure is helped by the fact that descriptions are standardized and contain a project code as well as the start- and the endpoint of the power line. These terms form our dictionary for detecting whether a power line is mentioned in a contribution. Footnote 3 63 percent of the contributions contain a reference to one or more power lines.

The lower panels of Table 4 give some information about the participation pattern: The majority of contributions (5626) come from citizens, Footnote 4 followed by local governments (603). The idea that open consulations generate more participation by citizens is thus supported by our data. However, citizen contributions are shorter than contributions by others actors. Most actor types write on average more than 1000 words, ecological associations being the most “wordy” with 1851 on average. Citizens, on average, contribute only 295 words, supporting the idea that they have less resources to draft contributions.

The phenomenon of interest is the emotional content of the consultation contributions. To measure this, we used the sentiment dictionary by Klinger et al (2016). The dictionary contains lists of words associated with seven emotions: anger, disgust, fear, joy, sadness, surprise and contempt. Other established German dictionaries code the overall positive or negative sentiment (Rauh, 2008), but the dictionary by Klinger et al (2016) allows us to assess the overall sentiment of the contributions, as well as the kind of emotion expressed.

Thus, the dictionary codes the thee “moral emotions” contempt, anger and disgust (Haidt, 2003; Rozin, et al., 1999). These three emotions are closely related in some conceptualizations (Brader & Marcus, 2013, p. 180). However, in terms of appraisal, disgust may reflect violations of notions of purity (as is plausible for the policy field at hand that concerns infrastructure construction in supposedly “pure” nature), contempt reflects violations of notions of proper behavior in a community, and anger reflects violations of personal autonomy (Rozin, et al., 1999). Sadness also has a negative valence, but is triggered if an actor perceives that she has no control over a situation (Smith & Ellsworth, 1985, p. 834). Fear, the last of the negative emotions, is characterized by high uncertainty about a situation (MacKuen, et al., 2010; Smith & Ellsworth, 1985, p. 834). Joy signifies feelings of pleasure and an overall positive mood (Brader & Marcus, 2013, p. 175), while surprise connotes the feeling of uncertainty.

We do not imply that this is an exhaustive list of emotions triggered by political phenomena, nor do we claim that these emotions are necessarily distinct; one of the empirical questions is whether their expressions are corelated. However, we have a plausible set of prototypical emotions that could in principle be related to infrastructure construction.

Table 1 shows examples of the dictionaries. Footnote 5 We applied this dictionary to the raw contributions, so the number we report is the percentage of words with emotional content in relation to the total number of words in a contribution.

Table 1 Excerpts from the emotions dictionaries
Full size table

The face validity of the dictionary coding is high, as many of the words used in the contributions are covered by the dictionary. Table 2 shows some examples. Some words appear in two dictionaries (e.g., frustration as anger or sadness), reflecting the fuzzy boundaries between the emotions. Thus, the dictionary may overestimate the emotional content of the contributions by double-counting words. On the other hand, Table 2 shows that some emotional words like beschämend (mortifying) are not part of the dictionary, so the emotional content might be underestimated. As our aim is not to assess the “true” emotional content of contributions, but to measure the emotional content of contributions in relation other contributions, this problem does not unduly distort our measurement.

Table 2 Example sentences from the contributions and their coding
Full size table

An open question is whether the emotional expressions in the contributions are distinct or whether they are correlated. Table 3 shows that the coding seems to address different forms of expression (even though some of the dictionaries overlap). The largest association is between fear and sadness; contributions that express fear also tend to express sadness. However, there is no high correlation between anger, disgust and contempt.

Table 3 Correlations between the codings of emotions
Full size table

Another open question concerns whether the dictionary is suited to code the sometimes formalized language of consultation contributions. There may be stylized ways of politely expressing dissent that the dictionaries cannot code correctly (hence, coding as joy what is only politeness). However, similar sentiment dictionaries have been applied to semi-formalized German political language documents like party manifestos and political speeches (Rauh, 2018), and we have a deeper look into some of the contributions to assess our coding.

To validate our dictionary and to make sense of the numbers, we used the dictionary to code three German reference texts. First, the documents by the TSOs in which they justify their power lines. We expect these documents to be “dry” and devoid of emotions. Second, Goethes “Sorrows of Young Werther.” This novel tells the story of an ill-fated love affair and final suicide of the protagonist. We expect this prime example of German romantic literature to be full of emotions. Third, we code an internet football fan forum after the club has won or lost against its archrival. Footnote 6 The results corroborate our expectations. The proportion of emotional terms in the TSO’s network development plan 2012 is 0.35 percent. Even the least emotional group of contributors uses three times as many emotional terms. Goethes Werther, on the other hand, contains 4.7% emotional terms and therefore more than double the proportion of emotional terms than the contributions we study. Moreover, if we disaggregate Werther into its constituting chapters and trace the pattern of emotions, we see that the emotion coding traces the flow of the story, with Werther oscillating between joy and sadness, but in the end being overcome by sadness and contempt for the world, and killing himself. The football fan fora also show a high level of emotions and clearly show more joy after beating the archrival. Footnote 7

Table 4 and Fig. 1 give an idea of the phenomenon we want to explain. We see that citizens’ contributions are more emotional than the contributions by all other actor types. On average, a contribution by a citizen contains 2.19 percent emotional terms, whereas a contribution by a state government contains only 1.12 percent emotional terms. The main emotion is joy, showing that most contributions contain conciliatory terms (e.g., “We welcome the opportunity to comment on…”), Footnote 8 followed by contempt, fear and sadness. Disgust does not play a major role, and surprisingly, anger is absent from the contributions, even though German newspaper commentary calls protesters “angry citizens” (Kostka & Anzinger, 2016). At least in the contributions to the consultation, this anger is not manifest.

Table 4 Emotional content of consultation contributions: mean percentage of words by emotion/mean number of words
Full size table
Fig. 1
figure 1

Emotional content in consultation contributions

Full size image

To conclude, our aim is to describe the emotional content of consultation contributions depending on the actor type contributing and on whether the contribution contains a reference to a specific policy. As the consultation is performed by both the TSOs and the FNA, and contributors might change their emotional tone depending whether they are addressing private or state actors, we use the consulting actor as a control. We do not claim to make a causal argument; rather, we use regression models as a way to describe patterns in the data. We include year dummies to account for unobserved heterogeneity over time, e.g., due to focusing events that increase or decrease the salience of the issue. Footnote 9 The estimation strategy is threefold: We first use simple OLS regression models. Second, we add the year dummies. Third, to account for the fact that emotional content only ranges from 0 to 100 percent, we estimate tobit models (reported in appendix).

The upside of our approach is the intersubjective replicability of the coding procedure: The dictionaries can be applied to all kinds of German texts relatively quickly. In comparison with political psychology, which often uses surveys (Marcus, et al., 2006), our approach has the advantage that we use textual data that was directly produced in the policy process. Analogous to the debate about hand coding versus automated coding of policy positions (Bunea et al., 2017), the downside of our approach is that we might miss some nuanced expressions of emotions that hand coding would reveal. However, as the first attempt to elucidate and describe the emotional side of consultations, our procedure should suffice.

The empirical application

According to our expectations, consultation contributions by citizens and contributions mentioning concrete policies should display more emotions than other contributions. Table 5 shows the results of OLS regression models that test these contentions. Model (1) uses the percentage of emotional content as the dependent variable, and models (2) to (8) split the emotional content into the seven emotions. The actor dummies indicate which kind of actor submitted a statement. Citizen being the baseline category, the coefficients of the actor dummies can be interpreted in relation to citizen contributions.

Table 5 OLS regressions: proportion of words signifying emotions
Full size table

Concerning the control variable, we see that it makes a difference whether contributions are submitted to the private TSOs or the regulatory agency FNA. The FNA receives more emotional content, in particular more “joy” (i.e., contributions are couched in friendlier terms), but also more fear and anger. In particular, the increase in anger fits well here: An antecedent of anger is perceived unfair treatment by an external agent (Brader & Marcus, 2013, p. 180). This dovetails with the fact that the FNA consultation is held after the TSO consultation each year. Previous research on the case shows that contributors use the FNA consultation to complain about the behavior of the TSOs (Fink & Ruffing, 2019).

Concerning the actor type, the result is that citizens’ contributions display more emotions than the contributions by most other actor types. Citizen is the baseline category for the actor dummies, and nearly all actor dummies are negative and significant. If we disaggregate the emotions, it is especially the prevalence of fear, sadness and to a lesser degree contempt that distinguishes citizen contributions from other contributions. Nearly all other actor types express less fear and sadness, and most other actor types express less contempt. Figure 2 illustrates this result in a comparison between a citizen’s contribution and the contribution by a state administration: ceteris paribus, the former will contain 2 percent emotional terms, the latter 1 percent emotional terms.

Fig. 2
figure 2

Predicted proportion of emotional terms. Based on model (1) in Table 5

Full size image

That citizens express more emotions might seem like a trivial result. However, this result nuances current scholarship that—based on interviews and focus groups—argues that many of the protesters against renewable energies in Germany are highly educated men with a background in the natural sciences, which use technical arguments (Zilles & Marg, 2022).

The result that citizens express more fear and sadness than corporate actors dovetails with results from political psychology. Sadness is often associated with a feeling of powerlessness (Smith & Ellsworth, 1985, p. 834), and fear arises if individuals see themselves as victims of overarching circumstances (Wagner, 2014). For both emotions, it is plausible that citizens express them more than corporate actors, as they are more likely to feel powerless and have the impression that grid construction is due to opaque political decisions.

Concerning our second expectation, the two regression strategies give different answers. The models without year dummies (Table 5) suggest that the overall proportion of emotional terms does not change if a concrete power line is mentioned. The models with year dummies (Table 6), however, suggest that the mentioning of a power line correlates with more emotional terms.

Table 6 OLS regressions: Proportion of words signifying emotions. Year dummies included
Full size table

However, both regressions point to the fact that the emotional profile of contributions changes if a concrete power line is mentioned: A contribution referring to a concrete power line contains ceteris paribus fewer expressions of joy and surprise,Footnote 10 but more expressions of fear, sadness and disgust. For example, when talking about the energy transition in general, contributors “begrüßen die Vorreiterrolle Deutschlands” (appreciate Germany’s leadership, joy). However, when referring to concrete power lines, contributors are “in Sorge um ihre Gesundheit” (fear for their health, fear), or see a “Gefahr für mich, meine Familie, die Tierwelt und die Landschaften” (danger for me, my family, the fauna and the landscapes, fear). They find the process “zermürbend” (backbreaking, sadness) and lament the “Wertverlust” of their houses (loss of value, sadness).

The increase in terms indicating disgust when concrete power lines are mentioned resonates with results from psychology: Disgust is elicited by violations of physical purity (Haidt, 2003). Hence, it is plausible to interpret the increase in terms expressing disgust as a reaction to the perceived impact of the power line on the environment (as opposed to abstract considerations of energy policy that do not have local consequences).

This emotional profile allows for predictions. Fear is seen as enhancing information-searching behavior. This would mean that citizens that refer to a concrete power line will search for more specific issue-based information (MacKuen, et al., 2010). This is plausible, as people have an incentive to search for information if a concrete power line is planned in their neighborhood. Disgust, on the other hand, is less researched. Most conceptualizations group it together with anger (Brader & Marcus, 2013, p. 180), which would mean that disgusted citizens are more likely to rely on heuristics, and motivated to join protests (Valentino, et al., 2011). Sadness, in turn, is a feeling of powerlessness and a lack of control over the situation (Smith & Ellsworth, 1985), which depresses political participation. Thus, the emotional profile of the contributions to concrete power lines may indicate different types of reactions: information-seeking about the policy, protest or apathy.

To make sense of the models, Table 7 shows the predicted proportion of emotional terms in relation to different constellations. The empirical mean proportion of emotional terms is 1.99 percent (see Table 4). The most emotional contributions are those by citizens addressing the FNA and referring to a concrete power line (2.36 percent emotional terms). This percentage may not seem high, but remember that a) the percentage is expressed as a relation to the raw, unprocessed text, and b) even Goethes Werther scores only 4.7 percent emotional terms.

Table 7 Predicted proportion of emotional words, depending on constellation of independent variables
Full size table

The high emotional content of contributions by scientific actors can be explained if we disaggregate the emotional content. Tables 8 and 9 do this for joy and contempt. We see that contributions by scientific actors express these two emotions often. This is due to the fact that contributions by scientific actors use friendly wording to contextualize their contribution (e.g., “we welcome the energy transition…”), but also use many of the words from the contempt dictionary to express criticism (false, erroneous, inappropriate and the like).

Table 8 Predicted proportion of emotional words denoting joy, depending on constellation of independent variables
Full size table
Table 9 Predicted proportion of emotional words denoting contempt, depending on constellation of independent variables
Full size table

The relatively large proportion of terms indicating joy is an open methodological problem for our approach. The dictionary coding applied to the whole contributions cannot disentangle whether joy is directed at the policy or “formal politeness.” Based on a close reading of the contributions and knowledge of the policy field, our interpretation is that there is a baseline of formal politeness. However, the result that less joy is expressed if a concrete power line is mentioned (Table 8) suggests that—at least for our case—joy varies systematically with the level of concreteness of a policy.

Table 10 finally looks at fear. Fear is the emotion that distinguishes citizens from the other actors. Most actors do not voice fear, but citizens do, the more so, if they mention a concrete power line. The words indicating fear are directly connected to the power line, the fear of electromagnetic fields, fear for the environment, fear for the loss of value of property, etc.

Table 10 Predicted proportion of emotional words denoting fear, depending on constellation of independent variables
Full size table

It is interesting to note that the dominant negative emotions of citizens are fear and sadness, but not anger. The literature argues that fear and anger are triggered by different processes and have different consequences: Anger is triggered if an individual perceives a normative violation by external actors, causes the use of routines and impedes information-searching behavior (Erisen et al., 2019). Fear is a reaction to uncertainty or the lack of resources to control a situation and “increases an individual’s propensity to seek out for new information about the threatening stimulus, deep and effortful processing of knowledge, and decreases reliance on long-standing political beliefs and partisanship” (Erisen, et al., 2019, p. 4). For public consultations, finding expressions of fear in the contributions thus might be good news (or at least better news than finding anger): Citizens participating might become more well informed and vigilant about the subject matter at hand. Furthermore, this may mean that making the procedures more transparent and offering more information could help to reduce fear.

To summarize, the results of our descriptive exploration are that emotions are an important part of consultation contributions and that citizens contributions contain more terms indicating emotions. The regression models with year dummies also suggest that contributions are more emotional if they mention concrete policies. Moreover, the emotional profile of contributions changes if concrete power lines are mentioned, with less joy and more fear, disgust and sadness expressed. Whether these emotional profiles have different behavioral consequences is an interesting question for further research.

Conclusion

Our article starts from the observation that the literature on public consultations often uses a resource exchange perspective and is less interested in the emotions displayed in consultation contributions. However, this perspective is lopsided, as emotions and cognition are intertwined. We argue that consultations can be analyzed with regard to the emotional content of consultation contributions, to elucidate the emotions that the proposed policies trigger. Using German electricity grid planning as a case, and a dictionary coding of consultation contributions, we test two expectations: First, citizen contributions contain more emotional terms than contributions by other actors. Second, contributions mentioning concrete policies (in our case: power lines) contain more emotional terms than contributions referring to the policy framework. Both expectation hold. Contributions by citizens contain more emotional terms than other contributions. Contributions mentioning power lines contain more terms indicating sadness, disgust and fear, but fewer terms indicating joy.

As a methodological caveat, we note that the coding procedure is no panacea, and interpretations of the percentages of emotional terms have to be made with reference to comparison texts and the policy context. In our case, consultation contributions seem to contain a baseline of formal politeness, resulting in much “joy.” Comparative studies of other contexts could elucidate how the dictionaries perform in other contexts (for example, parliamentary questions).

Our results have implications for several discussions. First, for the case of Germany, we corroborate the result that power line projects elicit fear in the population (Mueller, et al., 2019a; Zilles & Marg, 2022). Hence, acceptance of these power lines is low, even if the energy transition is judged favorably. Related, for the German system of interest intermediation, most research has focused on ministerial consultations and parliamentary hearings (Rasch et al., 2020). Our research suggests that if more open consultations are introduced in Germany, more citizens might participate—as research from other political systems already suggests (Rasmussen & Carroll, 2014, p. 449). However, including more citizens also means getting more emotional responses. An open question is how far the results from our case are generalizable. There are reasons for optimism: As discussed, the institutional setup of the German consultation is similar to the European Commission’s online consultations, and the main arguments about formal organizations vs. citizens and abstract vs. concrete parts of the policy are not specific to the German case.

Second, our approach offers a new perspective to assess consultation contributions. We show that many participants do not only voice policy positions, but also display emotions. An interesting research question is to connect our research to the research on framing. Current research, for example, simply codes on a five point scale where stakeholders lie between “fully against” a policy and “full support” (Eising & Spohr, 2017). As emotional expressions can also be used to express support or opposition, it seems worthwhile to research how a hand coding of support and opposition relates to the emotional content of contributions. Similarly, it would be interesting to elucidate whether emotional contributions contain less factual information, or whether emotional contributions contain the same amount of information, only presented in emotional language. If we follow political psychology, the “fearful” contributions should contain more information than the “angry” contributions.

Third, for public consultations as institutions, many authors argue that consultations are designed to attract the “right” kind of stakeholders and information (Braun & Busuioc, 2020), e.g., by strategically using participation rules. We contend that how well consultations work also depends on their ability to deal with the emotional content of contributions. We have shown what kind of emotions a public consultation elicits, and political psychology argues that different emotions have different antecedents and consequences. This suggests practical guidance for consultations. For example, if fear is the dominant emotion that shows up in the contributions, making the process more transparent might be the answer.

Fourth, our article shows a way to assess the emotions that a policy triggers. For many theories of public policy, the degree of conflict a policy generates is a central variable (Cairney & Weible, 2017, p. 62; Vogeler & Bandelow, 2018; Weible & Heikkila, 2017, p. 24). The current approach to assess the emotional quality of policies is to measure (social) media content, to conduct surveys or experiments (Maor, 2016, p. 203). Our proposal is to add dictionary coding of consultation contributions to the toolbox, as emotions tell us something about the degree of conflict in a policy field that is not captured by policy beliefs and preferences. An issue that is not yet covered in our approach is the interaction of emotions and policy preferences. Presumably, emotions like anger cause actors to overestimate the differences between policy preferences, and harden the boundaries between coalitions (Vogeler & Bandelow, 2018). However, that is a venue for further research that systematically studies the interaction between emotions and policy preferences over time.