Words and Attitudes of the Heart: The Emotional Content of Christian Nationalist Communications

Abstract

Christian nationalism has emerged as an important component of the relationship between religious identities and political attitudes. While several studies have analyzed the constellation of Christian nationalist elites and the effects of Christian nationalist orientations on public opinion, to date no study has explored how Christian nationalist elites message to the public or what effects these messages have. Moreover, the current literature lacks comparisons of Christian nationalism to other similar orientations. This study uses content analysis to compare the content and use of emotion language of Facebook messages of Christian nationalist, Christian Christian nationalism (CN) opposition, and patriotic groups. I find that these groups focus posts on issues that are stereotypical to the group identity, and that the use of emotion language differs by topic and group type. Additionally, groups’ use of emotion language shifts the emotional responses of readers, especially in Christian nationalist groups. This study adds to our understanding of the role of emotion in social media communications and the effects of social media communications on readers.

1. Introduction

In a 1994 speech, Pastor D. James Kennedy said:

Our job is to reclaim America for Christ, whatever the cost. As the vice regents of God, we are to exercise godly dominion and influence over our neighborhoods, our schools, our government, our literature and arts, our sports arenas, our entertainment media, our news media, our scientific endeavors—in short, over every aspect and institution of human society.

(Kennedy 2010, p. 127)

The idea that the United States is or should be a Christian country guided by Christian principles, also known as Christian nationalism, is a popular one in the United States, structuring public opinion on a range of issues (Davis 2018; Shortle and Gaddie 2015; Whitehead and Perry 2015, 2020). While some of these ideas are rooted in previous eras of American history (Fea 2018; Gorski 2017), what “godly dominion” looks like in any particular time or in response to any particular scenario is not clear. As a multivocal text and diverse tradition, the Bible and Christianity offer numerous, often contradictory lenses through which to view contemporary issues. How is the average Christian nationalist to inform their opinions?

Christian nationalism is an important aspect of the religious right, a network of non-governmental organizations, lobbying groups, and activists (Stewart 2020). These elites undertake the education and mobilization of the masses, defining for the public the priorities and preferences of a Christian America. This information is shared in many ways, but for day-to-day news, social media has become an important means through which Christian nationalist elites connect with the public (Freire 2014). On a day-to-day basis, what are Christian nationalist elites signaling? Which topics are most important to Christian nationalists, and how are Christian nationalist elites communicating their emotional responses to the events of the day? How are readers of Christian nationalist messages responding? And to what extent are any of these features unique to Christian nationalism, and not a function of the religious or patriotic elements of the group?

This article explores Christian nationalist social media messaging through a content analysis of Facebook posts from 2019. It compares the posts of Christian nationalist groups to Christian CN opposition and patriotic groups, finding that Christian nationalist social media messaging is distinct both in terms of the topics of interest but also in terms of the emotion communicated through posts. Moreover, this article concludes that readers, especially Christian nationalists, are responsive to the emotional content of messages.

2. Literature Review: Religious and Cultural Communication

In 1960, Klapper declared that communication produces, at best, minimal effects (Klapper 1960). In the years since, research has demonstrated that communication can influence opinion in significant ways, although effect size, causal mechanism, and moderating factors are vigorously debated (Tesler and Zaller 2017). In the field of religion and politics, most work on communication effects has centered on the influence of the local pastor, finding, in agreement with Klapper, limited effects. Not only are most pastors hesitant to engage in politicking from the pulpit (Djupe and Calfano 2013), but the one-hour-a-week sermon has a brief half-life (McClendon and Riedl 2019).

Instead, the congregational context and influence of lay leaders have emerged as more important shapers of opinion in the pews (Bean 2016; Djupe and Gilbert 2008). However, these explanations do not provide much leverage for understanding similarities across congregations, many of which operate outside denominational structures or are situated in denominations divided by the issues of the day (Crary 2021; Gordon 2020; Sherwood 2021).

Transcending individual congregations and denominations is a larger evangelical sub-culture, consisting of merchandising, publishing houses, music labels, radio stations, celebrity pastors, and interest groups. This evangelical sub-culture helps “individuals form bonds with other like-minded consumers, and these affinities form the basis of a shared cultural identity” as the “evangelical marketplace itself helps define who is inside and who is outside” (Kobes du Mez 2020, p. 9). This shared cultural identity is not simply based on religious identity, but uses the fusion of religious and national identities to lay claim to privileged status within the body politic (Whitehead and Perry 2020).

Despite increased attention being paid to the role of Christian nationalism in shaping attitudes and behaviors towards issues ranging from criminal justice reform to vote choice to public health behaviors (Davis 2018; Perry et al. 2021; Whitehead and Perry 2020), to date there has not been any examination of the communications content of influential movement leaders, nor of its effect on communications consumers. Moreover, much of the work on Christian nationalist elites is confined to those elites themselves (Fea 2018; Kobes du Mez 2020; Stewart 2020); the lack of comparison groups complicates efforts to draw clear causal connections between Christian nationalism itself and the outcomes of interest, as other elements of communication, such as the religious content of communication or moralistic frames, may contribute to attitude formation (Clifford 2019). Finally, emotional content plays a significant role in elucidating the frame of mind of communications’ authors and in shaping audience responses and subsequent opinion and behavior (Brader 2006; Valentino et al. 2011), yet there has, to date, been no analysis of the emotional content of Christian nationalist communication.

While much religious communication occurs through more traditional channels, such as talk radio, books, or religious television programming, social media serves an increasingly important role. Traditional Christian religious authority was vested in local clergy or members of the denominational hierarchy, but the Internet has ushered in an era in which traditional authority is challenged. Through social media, both traditional religious authority and alternative religious voices can develop a following, create community, and engage in dialog with followers. Religious discussions on social media are commonplace, especially for White evangelical and Black Protestants, and complements consumption of other religious media (Pew Research Center 2014). Traffic to religious sites is often driven by a desire to access faith-based information, particularly on Facebook (Brubaker and Haigh 2017).

3. Theory: Communication, Shared Identity, and Hypotheses

It is often assumed that religious groups are other-worldly focused, but in fact religious individuals and groups operate in ways that generate here-and-now benefits for themselves and attempt to provide benefits to supporters. In their analysis of the use of Twitter by influential evangelical leaders, Burge and Williams find that accounts use Twitter in ways that develop the “personal brand” of the owner (Burge and Williams 2019). Moreover, Christian organizations use social media communications in ways that are sensitive to their intended audience, reflecting the strategic underpinnings of religious communications (Wilson and Djupe 2020). Like non-religious groups, religious groups’ communication seeks to recruit and maintain membership.

One way in which groups seek to achieve influence over their membership is by cultivating identities. In his study of the National Rifle Association communication’s influence, Lacombe argues that “[a] politicized group identity enhances a group’s ability to influence politics via outside lobbying by increasing the political salience and intensity of emotions held among group members, make it easier to mobilize them to engage in various forms of political participation” (Lacombe 2019, p. 1344). Religious communications also contribute to identity cultivation by identifying in-groups and out-groups and linking them with norms and stereotypes (Hughes 2020). For example, in her ethnography of American evangelical churches, Bean describes how issue positions on topics like abortion are used as markers of identity (Bean 2016).

Christian nationalist rhetoric has long focused on issues related to the family and gender. In particular, abortion has emerged as the most important issue in politics among the religious right, which has considerable overlap with Christian nationalism (Bean 2016). During the 1990s, rising acceptance of same-sex marriage led the issue to become more important to Christian nationalists, who argued that the strength of the nation was dependent on the one-man-one-woman marriage model (Whitehead and Perry 2015). Transgender issues remain central to Christian nationalist lobbying efforts (Gabbatt 2021). Finally, religious liberty has emerged as a central frame, either on its own or when applied to other issues, as in the Supreme Court case Masterpiece Cakeshop, Ltd. V. Colorado Civil Rights Commission (Lewis 2017). But not all Christians connect their faith to political issues in the same way. Some Christian elites have become vocal advocates for the rejection of Christian nationalism. For Christian CN opposition groups, stereotypical issues include immigration and racial and economic justice (Wallis 2006, 2016).

Hypothesis 1.

Christian nationalists’ posts will have a heavier focus on abortion, LGBT issues, and religious liberty relative to Christian CN opposition posts. Christian CN opposition posts will focus on race, immigration, and economics relative to Christian nationalists’ posts.

Communications can not only build social identities, but also influence the emotions experienced by group members. Once individuals identify as part of a group, they begin to experience emotional reactions on the basis of their social identities (Cottrell and Neuberg 2005). There are three main emotions that are most commonly assessed in religion in communication: anxiety, anger, and disgust. Anxiety is a response to situations that seem uncertain or dangerous, like outbreaks of disease, terrorist attacks, or other events that are perceived to be life-threatening (Albertson and Gadarian 2015; Brader et al. 2010). The threat need not be associated with physical danger; for example, Brader found that images and music in campaign ads can trigger feelings of anxiety (Brader 2006), and Fea argues that the Christian nationalist movement is undergirded by fear (Fea 2018). Anger can result when goals seem to be slipping away but are still in reach, especially when blame can be assigned to another person or when the situation is deemed unfair (Lazarus 1991), or when violations of fairness or care are perceived (Russell and Giner-Sorolla 2011). Disgust is a less commonly studied emotion in political science. However, in psychology, disgust is often studied as an emotional reaction that protects humans from physical danger (for example, from decaying corpses), and more recently the study of disgust has been applied to symbolic or moral contamination, with some attention paid to the special role of disgust for religious individuals (Choma et al. 2016; Djupe et al. 2021; Miller et al. 2017). Finally, sadness is described as disappointment or a feeling of loss, and has been associated with slower cognitive processing, fewer blame attributions, and withdrawal (Kim and Cameron 2011).

As prototypical group members, the elites of organizations communicate their emotions around social identity-related stimuli through their communications. Readers who claim membership in the same social group respond to social identity-related stimuli, as well as taking emotional cues from the communications of elites, experiencing their own emotional reactions (Cottrell and Neuberg 2005; Lin and Haridakis 2017). Internet news consumption tends to be well-sorted, with individuals consuming news from like-minded sources. Homophilic information exposure and consumption should be especially strong on social media sites like Facebook, the algorithms for which funnel users to like-minded pages and in which individuals select pages to follow on a regular basis.

Hypothesis 2.

The use of emotion language in posts will vary by group type.

Hypothesis 3.

The type of emotion language in posts will vary by topic.

There is reason to believe that Christian nationalists, specifically, experience group emotions, such as a sense of threat (Green 2017). For example, the American Center for Law and Justice, headed by Jay Sekulow and one of the leading Christian nationalist legal advocacy firms, publicized their petition on their Facebook page on 10 May 2020. It states, “The Left is criminalizing the execution of the U.S. Constitution. The Deep State is undermining the rule of law and the conservative agenda. This is a constitutional moment for the American people. It’s a really big deal. It puts our republic at risk” (American Center for Law and Justice 2019). Posts like these communicate that there is an in-group (Christians, whose norms include being politically conservative and true Americans/constitutionalists) and an out-group (the Left, the Deep State) that poses an existential threat to the values of the nation (Bean 2016). It is probable that appeals like these lead to an emotional reaction, not because the average petition-signer is concerned that they will end up in court against the Deep State, but because the average petition-signer feels that there is a threat to their national or religious group (with these identities having been largely fused together for those in the Christian nationalist movement). Social Identity Complexity Theory would lead us to expect that these overlapping identities would lead to particularly strong effects (Roccas and Brewer 2002), especially as these identities are supported by strong organizational structures (Wilcox and Robinson 2011). On the other hand, Christian CN opponents are not as well-organized and experience more cross-cutting cleavages (Sitman 2021), which should dampen group emotional responses.

Hypothesis 4.

Emotion language in posts will correlate with reader reactions.

Hypothesis 4a.

All negative emotions will be negatively correlated with love emoji reactions.

Hypothesis 4b.

Anger language will be positively correlated with angry emoji reactions.

Hypothesis 4c.

Sad language will be positively correlated with sad emoji reactions.

Hypothesis 4d.

Readers of Christian nationalist posts will have stronger reactions to emotion language than readers of Christian CN opposition posts.

4. Methods

Content analysis provides a set of tools for characterizing message content, making inferences about message senders, and capturing reader response (Berelson 1952). I utilized Facebook posts because Facebook is the largest social media platform, and almost all organizations and individuals identified as central to the Christian nationalist movement use it as an informational and mobilizational tool, as evidenced by their robust presence on their social media sites. I compared three types of organizations: Christian nationalist, Christian CN opposition, and secular patriotic groups. A two-stage sampling strategy is used to identify (1) individuals and organizations that are prototypical of Christian nationalism and Christian non-nationalism; and (2) the specific texts to be analyzed.

Christian nationalist groups believe that Christianity should have a privileged role in the public sphere (Davis 2018; Shortle and Gaddie 2015; Whitehead and Perry 2015, 2020). They will argue that the principles they associate with Christianity should be enshrined in government policy. To identify prototypical Christian nationalist organizations, I turned to The Power Worshippers: Inside the Dangerous Rise of Religious Nationalism, in which Katherine Stewart identifies the major Christian nationalist organizations shaping the movement (Stewart 2020). These include think tanks that provide the ideological basis for the movement (WallBuilders, Chalcedon, Falkirk Center, and Family Research Council), legal advocacy firms (Alliance Defending Freedom, American Center for Law and Justice), and explicitly political organizations (Capital Ministries and Faith and Freedom Coalition). In Believe Me: The Evangelical Road to Donald Trump, John Fea developed the concept of the court evangelical, a religious leader who draws close to the presidency to maximize political power (Fea 2018). His book and blog highlighted the activities of court evangelicals, providing a sampling frame of powerful evangelical elite individuals. Among the names most frequently mentioned are Eric Metexas, Franklin Graham, Ralph Reed, and Robert Jeffress. All of these individuals and organizations have active public Facebook accounts.

Christian CN opponents are Christians who explicitly oppose Christian nationalism. Christian CN opponents, like Christian nationalists, root their positions within their religious beliefs. But they draw different conclusions about how Christianity should inspire politics. For example, instead of arguing that Christians should exercise dominion or heavy influence over government, Christian CN opponents are more likely to conclude that Christians should promote religious freedom and freedom of conscious. To identify Christian CN opposition groups, I turned to the principal supporters of the organization Christians against Christian Nationalism and to an open letter “Against the New Nationalism”, published in Commonweal (2019). Many of the signers were professors; since I am interested in the effects of messages on public opinion, I did not include them if they did not also have a strong public presence through Facebook or other media outlets. Christian CN opposition organizations included in the sample are legal organizations (Americans United for the Separation of Church and State), political organizations (Faith in Public Life, the Friends Committee on National Legislation, and the Baptist Joint Committee for Religious Liberty), think tank-type organizations (Red Letter Christians, Sojourners, National Council of Churches, Bonhoeffer Institute, Faithful America, and EthicsDaily), and prominent individuals (William J. Barber, Dr. Cornel West, Elizabeth Eaton, Michael Curry, Sister Simone Campbell, and Russell Moore).

Patriotic groups are non-governmental groups with missions that are patriotic in character. They have a national scope, meaning that they do not tie their activities to any specific religious or partisan group (Congressional Research Service 2020). To identify patriotic groups, I turned to the U.S. Code: Title 36. Patriotic and National Observances, Ceremonies, and Organizations (n.d., 36). I eliminated the organizations that specified religious orientations, age of membership, or occupational group. Of the remaining groups, I identified the groups that would have broad reach, resulting in the selection of American Gold Star Mothers, the Congressional Medal of Honor Society, Daughters of the American Revolution, and Veterans of Foreign Wars.

These groups provide important variations in the type of messaging along two dimensions. First, Christian nationalist and Christian CN opposition groups differ in their beliefs about the role of Christianity, specifically in the public sphere. Christian nationalists seek a broader role in which legislation and national symbols are at a minimum influenced by Christianity, while Christian CN opponents advocate for less Christian presence in government. The inclusion of patriotic groups also allows for the exploration of the extent to which religious framings themselves alter the message format and reader reaction.

Having identified the organizations and individuals most representative of the Christian nationalist, non-Christian nationalist, and patriotic leadership, I used CrowdTangle1 to cull the text of all posts from each of these pages from 2019, resulting in a population of 25,933 posts (CrowdTangle Team 2021). I randomly sampled 1600 posts, deleting any duplicates for a final sample of 1558 posts. 2019 was selected as the year for analysis because it was prior to the 2020 election cycle and COVID-19 pandemic, both of which events would likely shift emotional language and response, thereby limiting the generalizability of the project.

After the posts were sampled, common stop words, punctuation, and capitalization were removed. The NRC Emotion Lexicon was then applied to the resulting list of words, providing a count of each type of emotion word (anger, anxiety/fear, anticipation, trust, surprise, sadness, joy, disgust) used in each post. The NRC Emotion lexicon was developed and tested through crowdsourcing and has become a well-established lexicon for emotions research (Mohammad and Turney 2010). The word counts were then converted into percentages such that the emotion scores represent the percent of emotion words related to each emotion.

To measure readers’ reactions, I used the data provided by CrowdTangle on readers’ emoji reactions (Like, Love, Sad, Angry, Haha, Care, Wow). “Like” was the most common reaction by far, but its meaning is ambiguous. Some readers click “Like” to acknowledge the post, others to signal agreement. Consistent with other analysis (Eberl et al. 2020), I have not included “Like” emoji reactions in my analysis as a marker for any particular kind of emotion. Using the remaining emoji reactions, I calculated the share of reader emoji reactions for each emoji type. For example, if a post has an angry reader emoji score of 37, 37% of the reader emoji reactions were Angry.

Posts were also coded by topic. Topics in some posts were intertwined in such a way that coding for only one topic would lead to unreliable and misleading results. For example, there was a series of posts about whether religious healthcare providers should be required to provide contraception that they believe causes abortion, and thereby violate their religious beliefs. These posts are about abortion, but presented in a religious liberty frame. I decided to allow up to 2 codes for each post. Of the 1382 emotion posts, 319 received 2 codes. As a check on topic coding, a second coder was given a random sample of 175 posts (over 10% of the sample) and asked to code them by topic using a codebook provided by the author. Intercoder percent agreement was 74.9% and Krippendorff’s alpha was 0.687 (Freelon 2010).

Finally, posts were also coded for partisan references. I created a dummy for any mention of President Trump (including the words “Trump” or “president” if used in references to President Trump); a dummy for mentions of non-Trump Republican politicians and politics (including specific politicians like McConnell, “conservative” in reference to political ideology, or “Republican” in reference to the political party); and a dummy for mentions of Democratic politicians and politics (such as individuals like Pelosi or Biden, “Democrat” in reference to the political party, or “liberal” in reference to the political ideology).

5. Findings

5.1. Stereotypicality and Topic Analysis

Hypotheses 1 concerned post topics an organization, arguing that topics more central to a group’s identity would appear in posts more often.

Table 1. Percent of posts addressing each topic, by organization type.

	Christian Nationalists	Patriotic	Christian CN Opponents
Abortion	106 14.46%	0 0.00%	3 0.42%
Race and Ethnicity	11 1.50%	0 0.00%	108 15.25%
Immigration	7 0.95%	0 0.00%	72 10.17%
Military Affairs	6 0.82%	16 13.68%	2 0.28%
Criminal Justice	5 0.68%	0 0.00%	34 4.80%
Education	31 4.23%	0 0.00	11 1.55%
Economics	5 0.68%	0 0.00%	47 6.64%
Ideologies and Political Behavior	93 12.69%	3 2.56%	47 6.64%
International issues	63 8.59%	0 0.00%	45 6.36%
Women	6 0.82%	0 0.00%	26 3.67%
Announcements	165 22.51%	31 26.50%	102 14.41%
Institutions	22 3.00%	0 0.00%	22 3.11%
Health Care	3 0.41%	2 1.71%	6 0.85%
Religious Liberty	111 15.14%	0 0.00%	35 4.94%C2
LGBT Issues	39 5.32%	0 0.00%	26 3.67%
Environment	1 0.14%	0 0.00%	29 4.10%
Guns	8 1.09%	0 0.00%	21 2.97%
Inspiration	226 30.83%	68 58.12%	155 21.89%
Evolution/Science	5 0.068%	0 0.00%	0 0.00%

Notes: Bold results are those that are over 5%.

presents the topics that were coded, providing the percent of posts falling into each category. Because some posts address multiple topics, the percentages for each organization will not sum to 100. Inspirational posts are the most common topic for all groups. These posts may provide a Bible verse, a tribute to an inspiring person, or a motivational comment, but they do not contain political content. This finding is consistent with other studies that have found that even politically active religious actors retain a focus on religious (or, in the case of patriotic groups, generally inspiring) topics (Burge and Williams 2019). Announcements (such as advertisements for book sales or notices about upcoming streamed worship services) constituted the other important non-political category for all groups. But then the topics groups posted on diverged significantly. Among Christian nationalists, the most common political topics were religious liberty (15.14%), abortion (14.46%), and ideologies and political behavior (explicit endorsement of ideological thinking or political mobilization) (12.69%). LGBT-related posts constituted 5.32% of posts, a significant number, but reflecting a declining focus on LGBT issues since Obergefell.

The distribution of topics for the Christian CN opposition pages also points toward confirmation of the theory that communications stress stereotypical topics. Inspiration posts and announcements again constitute a large percentage of the posts (21.89% and 14.41%, respectively). Very few of the posts focus on abortion or religious liberty, suggesting that Christian CN opponents have their own agendas and are not simply responding to Christian nationalists or the religious right. Race and ethnicity posts are the most common political post (15.25% of all posts), with immigration being the second most common topic (10.17%). Economics-related posts represent 6.64% of the total. Other important topics are ideologies/political behavior and international issues. For patriotic groups, military affairs topics are most common (13.68% of posts). Instead of simply responding to each other, each group is emphasizing the topics it feels are central to its own identity, as expected by Hypothesis 1.

5.2. Emotion as Signaled in Posts

Hypothesis 2 concerns the use of emotion language in posts, arguing that different types of organizations should manifest different emotional profiles. Figure 1 shows the percent of each kind of emotion language found in posts. The mean of sadness language is not significantly different across the groups. Disgust language comprises 5.52% and 4.80% of Christian nationalist and Christian CN opponent emotion language, respectively, and 2.47% of emotion language in patriotic groups. The difference between patriotic groups on one hand, and the two religious groups on the other, is statistically significant.

The groups also differ in their use of fear language; 14.16% of Christian nationalists’ emotion language reflects fear, compared to 10.76% for patriotic groups and 11.43% for Christian CN opposition groups. T-tests show that Christian nationalists use significantly more fear language than Christian CN opposition (p = 0.003) and patriotic groups (p = 0.042), providing support for the argument that Christian nationalist elites communicate fear at a high rate.

These data allow for ranking the types of emotions communicated by each group.

Table 2. Means of percent of emotion language by organization type.

	Means (Standard Error)
	Christian Nationalist	Patriotic	Christian CN Opponents
Disgust	5.516 (0.386)	2.474 (0.637)	4.803 (0.327)
Anger	7.693 (0.457)	5.473 (1.065)	9.826 (0.590)
Fear	14.164 (0.621)	10.762 (1.710)	11.426 (0.604)
Sadness	6.824 (0.446)	6.332 (1.119)	7.820 (0.484)

shows that, of these three emotions, Christian nationalist posts utilize fear language more than anger language, which is more common than disgust language. T-tests indicate that fear language is greater than anger language (t = 9.401, p = 0.000), fear language is greater than disgust language (t = 12.349, p = 0.000), and anger language is greater than disgust language (t = 3.916, p = 0.000). Sadness language is not statistically indistinguishable from anger or disgust but is less common than fear language (t = 10.334, p = 0.000).

Finally,

Table 3. Effect of organization type on emotion language use.

	Disgust	Sadness	Fear	Anger
Christian nationalist	2.701 ** (0.954)	0.297 (1.264)	3.457 * (1.680)	2.067 (1.422)
Christian CN opponent	1.939 * (0.953)	1.301 (1.263)	0.828 (1.678)	4.210 ** (1.421)

Notes: Coefficients are estimates from OLS; standard errors are in parentheses. Patriotic groups are the reference group. Significance levels: ** < 0.01; * < 0.05.

provides further support that these groups use emotion language differently, comparing Christian nationalist and Christian CN opposition posts to patriotic posts. Posts from Christian nationalists and Christian CN opponents are more likely than patriotic groups to use disgust language. Christian nationalist posts have, on average, 3.457% more fear language than patriotic posts, and Christian CN opposition posts have, on average, 4.210% more anger language than patriotic posts.

Hypothesis 3 proposes that the topic should influence the type of emotion language used. I find some support for this hypothesis. Topics that were addressed by at least 5% of posts for each group type are analyzed. First, I conducted t-tests to compare the means of each kind of emotion language between posts that do and do not discuss a particular topic. For Christian nationalists, there were no statistically significant differences in emotion language for the LGBT and ideology topics; for Christian CN opponents, there were no statistically significant differences in the race/ethnicity, ideology, and religious liberty topics. The bar graphs presented in Figure 2 show the mean of each type of emotion language, comparing across topics. Christian nationalist posts about abortion use significantly more disgust, sadness, and fear language. Christian nationalist posts about religious liberty and international affairs use significantly more anger language. Patriotic groups’ posts about military affairs use significantly more sadness language. Christian CN opposition posts about immigration and international affairs use more fear language; posts about economics use more sadness language, and posts about international affairs use more anger language. Otherwise, the differences in type of language use are not statistically significant. Groups are not using the same type of language when writing about diverse topics, and the topics that are most central to a group’s identity (especially abortion for Christian nationalists) use language that is more emotion-laden than posts that are about other topics.

These findings are confirmed by a larger OLS regression model that measures the effect of each type of topic while holding all the others constant (presented in

Table 4. The effect of topic on emotion language for Christian nationalists.

Christian Nationalists Only	(1)	(2)	(3)	(4)
	Disgust	Sadness	Anger	Fear
Abortion	7.197 ***	7.453 ***	−4.115 **	2.682
	(1.486)	(1.720)	(1.742)	(2.381)
Guns	−1.702	2.559	1.781	2.489
	(3.957)	(4.578)	(4.638)	(6.337)
Environment	−4.181	−5.578	−8.239	−11.435
	(9.492)	(10.982)	(11.126)	(15.203)
LGBT	−1.495	−2.389	−2.872	−2.157
	(1.956)	(2.263)	(2.293)	(3.133)
Immigration	−4.163	−0.370	8.110	22.821 ***
	(4.330)	(5.010)	(5.075)	(6.935)
Religious Liberty	0.122	1.107	5.473 ***	−0.595
	(1.221)	(1.412)	(1.431)	(1.955)
Race and Ethnicity	8.117 **	1.726	−1.745	−7.885
	(3.148)	(3.642)	(3.690)	(5.042)
Healthcare	0.431	1.588	1.914	−5.555
	(5.473)	(6.332)	(6.415)	(8.766)
Institutions	2.987	−0.650	3.520	4.060
	(2.388)	(2.763)	(2.799)	(3.825)
Women	−2.557	−1.759	−1.407	12.663 *
	(4.438)	(5.134)	(5.202)	(7.108)
Economy	1.481	3.115	−8.984 *	16.296 **
	(4.333)	(5.014)	(5.079)	(6.941)
International	1.408	1.861	0.841	0.048
	(1.567)	(1.813)	(1.837)	(2.510)
Ideology	−0.496	−1.668	−2.257	0.037
	(1.413)	(1.634)	(1.656)	(2.263)
Evolution	−4.331	−6.878	−6.089	−13.558
	(5.473)	(6.332)	(6.415)	(8.766)
Military	8.793 **	−3.247	3.457	12.682 **
	(3.870)	(4.477)	(4.536)	(6.198)
Education	−0.448	−3.500	−2.245	−2.694
	(2.100)	(2.430)	(2.461)	(3.363)
Criminal justice	−0.471	−3.469	9.383*	−2.598
	(4.731)	(5.473)	(5.545)	(7.577)
Non-Political	−0.461	−2.590 *	−5.612 ***	−8.329 ***
	(1.267)	(1.466)	(1.485)	(2.029)
Democrat mention	5.113 ***	3.963 *	3.246	−1.027
	(1.907)	(2.206)	(2.235)	(3.054)
Trump mention	−0.419	−0.713	−4.802 ***	−7.872 ***
	(1.466)	(1.696)	(1.718)	(2.348)
Republican mention	−5.446 **	−3.816	1.917	−1.849
	(2.429)	(2.811)	(2.847)	(3.891)
Constant	4.331 ***	6.878 ***	10.255 ***	17.725 ***
	(1.219)	(1.410)	(1.429)	(1.952)
Observations	634	634	634	634
R-squared	0.121	0.118	0.136	0.129

Notes: Coefficients are estimates from OLS; standard errors are in parentheses. Significance levels: *** < 0.001; ** < 0.01; * < 0.05. Bold results are those that are statistically-significant.

). Over the past 40 years, religious groups have sorted along partisan lines, with conservative Christians becoming more affiliated with the Republican Party and the religious left more associated with the Democratic Party (McCarthy et al. 2019). Consequently, the models control for mentions of partisan actors, which, on their own, may explain emotion language. Because emotion shifted more significantly for Christian nationalists, the model below assesses only Christian nationalist posts. Here, abortion is associated with more disgust language and sadness language, but less anger language. Religious liberty posts were associated with anger language, much more so than fear language. A few of the results for other topics are worthy of mention. Race and ethnicity topics were associated with more disgust, but upon closer examination these posts discuss, with disgust language, instances of racial discrimination. Finally, immigration and economy topics have strong effects on anger and fear, as demonstrated by large coefficients. However, the small number of posts addressing either of these topics should limit our confidence in these results.

Groups create and maintain social media pages to communicate with readers and, if possible, influence readers. Hypothesis 4 projects that the emotion language used by groups will affect the emotional reaction of readers. OLS regression was used to calculate the effect sizes for each type of language on reader emoji reactions for Christian nationalists and Christian CN opponents; the results are presented in

Table 5. Effect of emotion language on emoji reader reactions, by organization type.

	Christian Nationalists			Christian CN Opponent
	(1)	(2)	(3)	(4)	(5)	(6)
Language Use	Love Emojis	Angry Emojis	Sad Emojis	Love Emojis	Angry Emojis	Sad Emojis
Anger	−0.560 ***	0.221 **	0.240 **	−0.241 *	0.060	0.087
	(0.134)	(0.074)	(0.082)	(0.121)	(0.060)	(0.091)
Disgust	−0.455 **	0.308 ***	0.060	0.172	0.088	−0.251
	(0.163)	(0.090)	(0.100)	(0.220)	(0.110)	(0.166)
Fear	−0.297 **	0.113 *	0.194 **	−0.497 ***	0.105	0.218 *
	(0.099)	(0.055)	(0.061)	(0.113)	(0.056)	(0.085)
Sad	−0.658 ***	0.267 ***	0.388 ***	−0.552 ***	0.041	0.530 ***
	(0.143)	(0.079)	(0.088)	(0.137)	(0.068)	(0.103)
Constant	79.504 ***	4.182 **	7.249 ***	75.230 ***	6.116 ***	14.193 ***
	(2.350)	(1.300)	(1.437)	(2.658)	(1.327)	(2.000)
Observations	607	607	607	515	515	515
R-squared	0.133	0.092	0.089	0.088	0.015	0.072

Notes: Coefficients are estimates from OLS; standard errors are in parentheses. Significance levels: *** < 0.001; ** < 0.01; * < 0.05.

.

First, Hypothesis 4a anticipates that “love” reader emoji reactions will be negatively correlated with all four types of negative emotion language (anger, disgust, fear, and sadness). There is fairly consistent support for this hypothesis. For Christian nationalists, increases in the use of negative language decrease the use of love emojis. For every percent increase in the use of anger language, for example, the percent of emoji responses that are “love” decrease by 0.560. These results are all significant at the 0.01 or 0.001 level, and sad language use has a larger effect on “love” reactions than on any other type. For Christian CN opponents, the use of anger, fear, and sad language also decrease “love” reader reactions. Disgust language does not have a statistically significant effect on “love” emoji reactions. In comparing the two group types, anger does not depress “love” reactions for Christian CN opponents as much as it does for Christian nationalists—the coefficient for Christian CN opponents is half the size of the coefficient for Christian CN opponents. On the other hand, fear language has a larger effect on Christian CN opponents than it does for Christian nationalists. Disgust language only suppresses “love” reactions for Christian nationalists, and the effect of sad language is only slightly larger for Christian nationalists. These results support the argument that negative emotion language depresses “love” reactions.

Hypothesis 4b argues that anger language should be positively correlated with angry emoji reactions. For Christian nationalists, there is a positive relationship between anger language and angry emoji reactions, and this relationship is statistically significant. However, the three other emotions also contribute to angry emoji reactions, and the effects of disgust and sad language are stronger than those of anger language. For Christian CN opponents, the sign of the effect of anger language is positive, but the coefficient is small and does not achieve statistical significance. No negative emotion language produces statistically significant results. Hypothesis 4b is supported for Christian nationalists, but not for Christian CN opponents.

Hypothesis 4c argues that sad language should be positively correlated with sad emoji reactions. For Christian nationalists, the coefficient for sad language is positive and is statistically significant. Anger and fear language also increase sad emoji reactions, but the coefficient size is smaller than for sad language. For Christian CN opponents, sad language also produces a statistically significant and positive effect on sad emoji reactions. Fear language also increases sad emoji reactions, but sad language has a stronger effect. For both Christian nationalists and Christian CN opponents, then, sad language produces sad reactions.

Finally, Hypothesis 4d predicts that, because Christian nationalists have more overlapping identities than Christian CN opponents, Christian nationalists should have stronger responses to emotion language than Christian CN opponents. In terms of the number of results that are statistically significant, there is some support for this hypothesis. Eleven of the coefficients for types of emotion language are statistically significant for Christian nationalists, while only five of the coefficients are significant for Christian CN opponents. Comparing the models across the two groups, the r-squared value for the Christian nationalist model is larger than the r-squared for the Christian CN opponent model. For Christian nationalists, emotion language explains 13.3% of the variance in love emoji reactions; for Christian CN opponents, emotion language explains 8.8% of the variance. For Christian nationalists, emotion language explains 9.2% of the variance in angry emoji reactions; for Christian CN opponents, emotion language explains only 1.5% of the variance. And for Christian nationalists, emotion language explains 8.9% of the variance in sad emoji reactions, while for Christian CN opponents, emotion language explains only 7.2%. Both the larger number of significant results and the larger r-squared values for Christian nationalist models point to stronger reactions among Christian nationalists than Christian CN opponents, supportive of Hypothesis 4d.

6. Discussion

Messages matter. Social media communications are, in part, about building identities—about signaling the topics that are important to the group and about communicating how group members feel about those topics. The three groups studied here have some elements in common: Christian nationalist and Christian CN opposition groups are both religiously oriented and both engage in discussion of political topics, while Christian nationalist and patriotic groups are both associated with strong support for the United States. However, these groups work to develop distinct identities, focusing their social media presence on stereotypical topics that are unique to each group type. Christian nationalists post most frequently on abortion, religious liberty, and political ideologies; Christian CN opponents post on race, ethnicity, immigration, and economic issues, and patriotic groups post on military issues. Moreover, they communicate different emotional profiles. Christian nationalist posts have a higher concentration of fear language, while Christian CN opponents’ posts stand out for high levels of anger language. Message topics are also related to the types of emotion language used.

The use of emotion language in turn influences reader reactions. Negative emotion language is strongly correlated with fewer “love” reader reactions. The relationship between specific emotions (sad language and sad emoji reactions; anger language and angry emoji reactions) is also present, although other emotion language contributes to these emoji reactions as well. Finally, the relationship between emotion language and reader response is stronger for Christian nationalists than for Christian CN opponents.

This study illustrates several implications for future research. First, the emotional content of language has distinct profiles, such that emotion language profiles can vary by topic or group type, and different types of emotion language have different effects on readers. Simple valence measures of emotion (positive or negative) are a starting point but lack the ability to make more nuanced arguments about emotion. Second, comparative perspectives are helpful. Christian nationalism has become a popular topic over the past few years, but most studies limit their analysis to Christian nationalism itself. Comparing Christian nationalism to other forms of politicized religion (like the religious left) and to groups with a strong nationalistic bent (such as patriotic groups) draws our attention to the unique attributes of Christian nationalism itself. For example, even though past work has drawn attention to punitive attitudes or support for authoritarianism (Davis 2018; Gorski 2020), both of which are rooted in anger over lost privilege and feelings of persecution, this study finds that anger is actually more prevalent on the Christian left. This raises questions about how much of Christian nationalist impulses are driven by anger over trespassed symbolic boundaries versus moral certitude that accompanies politicized religion.

This study is a first exploration of Christian nationalist social media messaging, and therefore paves the way for future work. First, a more extensive sample of posts should be analyzed. Future work should capture a larger number of organizations across a more nuanced array of Christian nationalism support and of secular vs. religious patriotism/nationalism. Additionally, in this iteration, topics were coded by hand, a labor-intensive process that limited sample size, both in terms of the number of posts sampled and the number of groups included in the study. Future work should use computerized topic modeling to identify key topics areas, and a comparison of computerized topic analysis to the more traditional coding approach developed here would be particularly beneficial. Second, while this study focused on negative emotion language, future work should examine the roles of more positive emotion language, such as trust or joy. These positive emotions may play important roles in building communities, especially imagined communities in online spaces, in mobilizing for political action, and in generating interactions with social media posts. Finally, more attention should be paid to religious liberties framings. On the one hand, some research points to the conclusion that religious liberties framings increase tolerance (Lewis 2017), making this type of communication a positive one for promoting intergroup interactions. On the other hand, though, this study finds that religious liberty topics are associated with anger, and other authors note that feelings of persecution may be contributing to fear and anger among Christian nationalists (Gorski 2020).

A recent surge of literature has drawn attention to the role of religion, particularly Christian nationalism, in shaping the public’s attitudes. This article has contributed to that literature by elucidating the role of social media in drawing boundaries around group identities and in describing the type of emotional cues provided, as well as by linking elite messaging to public responses. Given the high likelihood that religion will continue to influence public attitudes, further research into these links provides helpful insights.