The Dimensions of Political Speech – Conceptualising the Origins of Political Hostility

Daniel Knoth, University of Warwick

Abstract

This paper introduces a novel theoretical framework to analyse the increasing polarisation and hostility within US politics. The framework categorises political speech into three interrelated dimensions: locutionary, perlocutionary and illocutionary. By conducting an analytical review of the literature on political behaviour and polarisation and conceptualising the debate within this framework, the paper concludes that the increasing hostility is predominantly influenced by transformations within the illocutionary dimension rather than by changes occurring within the locutionary one.

Keywords: Speech Act Theory, Growing Political Hostility, Conceptualising Polarisation, Issue Versus Identity Polarisation, Conceptualising Partisan Behaviour

Introduction

The US political climate has grown increasingly hostile over the recent decades. Compared to 1960, when fewer than 5 per cent of Republican or Democratic parents expressed negative feelings towards the idea of their children marrying a member of the opposing political party, by 2008, this had grown to 25 per cent of parents, and from 2012, more than half of US parents felt like this (Iyengar et al., 2012). This animosity extends far across various areas of life, affecting social connections, hiring decisions and college applications (Amira et al., 2021; Iyengar et al., 2019; Overgaard, 2024). Reasons for this rise in political hostility are contested within the literature, with explanations ranging from a changing media environment (Steppat et al., 2022), biased exposure to news (Spohr, 2017) and the alignment of partisan identities (Mason, 2016) to issue polarisation driven by party elites (Jang et al., 2024).

This paper aims to clarify this debate by introducing a new theoretical framework that distinguishes between three different dimensions of political speech. By conducting an analytical review of the literature, I argue that conceptualising the debate within this framework can elucidate the area where political behaviour changes occur. I will begin by introducing the paper’s theoretical framework based on Austin’s speech act theory. I then briefly introduce the theoretical literature on political behaviour and explain how I incorporated this into my framework before discussing the empirical literature and different explanations of the changing political environment. Although this phenomenon is not unique to the United States, with similar changes in political behaviour occurring across Europe (Reiljan, 2020; Wagner, 2024), this paper will focus on the US as it has a broader and more comparable range of empirical evidence.

Analytical framework

In ‘How to Do Things with Words’, J. L. Austin argues for a distinction between three different so-called ‘speech acts’ that are present within our everyday speech (Austin, 1975). He defines the locutionary act as the literal utterances of our words and their immediate content. The perlocutionary act is defined as the consequences caused by speech, and finally, the illocutionary act consists of the meaning contained in speech (Austin, 1975). This speech act theory provides an analytical framework that allows us to conceptualise and understand different aspects of our language beyond the surface level of communication. While other applications of Austin’s framework within the political literature have commonly used it as a definitional framework of what falls under political speech (Harris and McKinney, 2020), such as in Langton’s prominent discussion of pornography (Langton, 1993), this paper takes a different approach. Instead of using Austin’s theory to define speech, this paper will build on the analytical framework it offers to distinguish between and conceptualise three different types of political speech acts. I use the term ‘political speech act’ in the broadest sense, referring to all forms of political expression, drawing on Estlund, who argues that our political actions, such as voting, are political speech acts due to their advocative and expressive nature (Estlund, 1990). Consequently, within this framework, all political expressions, such as wearing a ‘Make America Great Again’ cap or placing a ‘Vote for Biden’ sign in front of one’s house, are considered political speech acts. Therefore, given the analytical premise that these acts can be considered speech acts, this paper focuses on conceptualising different types of political speech, where, paralleling the Austin framework, it distinguishes between three different dimensions of political speech. The locutionary dimension here contains the literal content of a political issue position. This can be any economic or normative interest position, such as ‘a free market is best for the economy’ or ‘Abortion should be legal’. The perlocutionary dimension will contain all the actions one performs by expressing one’s political beliefs, such as voting in an election, demonstrating against a new policy or engaging in debate with others on political issues. Finally, the illocutionary dimension will contain all the personal meaning contained within political speech, such as thinking, ‘My team won!’ in response to one’s preferred presidential candidate winning the election or identifying as a member of a political group by saying, ‘I am a conservative’.

In this paper, I aim to reconceptualise the debate surrounding polarisation and hostility in the political climate. To this end, I will first introduce the two most prominent theoretical camps, discussing voting behaviour and political alignment within the literature, before mapping them onto my framework.

Theoretical models of voting behaviour

There are two broad groups of theories on voting behaviour and political alignment within the theoretical literature. I will focus my examples on voting as this has the broadest overlap between theories; however, these theories are also used to explain political behaviour more generally, of which voting is one expression. The two groups of theories differ in the factors they ascribe to influencing the voter’s choice. The first group of theories explains voting as a utility-maximising process where voters choose the party they believe will best satisfy their issue preferences. The two most prominent theories describing voting choices as ‘issue-based’ are the spatial and the valence models. The former, primarily influenced by Downs (1957), supposes that voters can locate themselves on one or more policy dimensions and will vote for the party closest to their preferences on this spatial dimension. If we consider a classical left-right spectrum, voters will simply vote for the party closest to their position on this spectrum. The latter model, based on a critique of the Downsian model by Stokes, argues that electoral decision-making is based on ‘valence issues’ (Stokes, 1963: 373), meaning that voters not only decide based on policy but also on which party or candidate is best suited to enact those policies – for example, based on the evaluation of the candidate’s qualifications or the parties past performances (Sanders et al., 2011). However, the unifying characteristic of these theories is that they conceptualise voters as making their choices based on a rational consideration of their policy interest and an issue-oriented decision on which party will maximise it.

The second group of theories, on the other hand, argues for the influence of social ties and identities on political behaviour. The sociological explanation of voting behaviour supposes that voting is driven by social pressure and that it is influenced by ‘group influences exerted upon them’ within their community (Lazarsfeld, 1944: 330). On the other hand, psychological theories argue that it is not social pressure but psychological attachment to different political groups and parties that influence our voting choices (Achen and Bartels, 2016; Campbell et al., 1960). Contrary to the ‘issue’ theories, where partisanship is merely the vessel for issue beliefs or represents a ‘running tally’ of past performance (Bartels, 2002), the psychological theories instead argue that it reflects an internal judgement of what ‘people like them’ believe (Achen and Bartels, 2016: 266). Group loyalties, therefore, matter far more than the actual policy details, and people will thus adopt the views that match whatever their identity aligns with (Achen and Bartels, 2016: 231).

Both groups of theories explain factors that influence political behaviour. The first group argues that it is largely issues that affect it, whereas the latter emphasises the channel of identity. Mapping this within my theoretical framework, we can draw out the three dimensions here. Observable political actions, such as voting, demonstrating and disagreeing with others in a debate, all fall within the perlocutionary dimension. Issue beliefs fall into the locutionary dimension, and identity considerations and attachments fall into the illocutionary dimension. However, the locutionary and illocutionary dimensions are not readily evident when it comes to observation. As outside observers, we only see the perlocutionary acts that an individual performs. For example, we can observe someone wearing a ‘Make America Great Again’ cap. However, from this observation alone, we cannot infer the other dimensions of political speech that this perlocutionary act expresses. For example, the perlocutionary act of supporting the Trump campaign by wearing this cap might be motivated by locutionary factors, such as a belief that Trump’s proposed policy to decrease the corporate tax rate will be beneficial for small businesses, or by illocutionary factors, such as an identification with a particular idea of US society and values that Trump represents for them. Therefore, we can think of every perlocutionary act to be influenced by factors from the locutionary or the illocutionary dimension. For this, I propose to imagine the three political dimensions as three layers, with the locutionary one at the bottom, the perlocutionary dimension in the middle and the illocutionary one at the top. Now, the issue-centred theories argue for a bottom-up approach, where perlocutionary acts are primarily influenced by the lower locutionary dimension, whereas the identity-centred theories emphasise a top-down approach where perlocutions are primarily driven by illocutionary considerations. However, as both political issues and the human decision-making process are immensely complex, it is probably more plausible to think of every perlocutionary act being influenced by factors from both the locutionary and the illocutionary dimensions to varying extents. For example, one might feel more connected to the identity of the Republican Party but think strongly about a specific issue position, such as abortion, that goes against the party line. Both the top and bottom layers are simultaneously influencing political behaviour. For the purpose of this article, we do not need a definitive answer to the debate within the theoretical literature about how strong the issue channels and the identity channel are compared to each other. Observing a change within one dimension while the other remains constant would suffice to determine the area from which the shift in political behaviour originates without determining the absolute influence issues or identities have on our political behaviour.

In the following section, I will engage with one of the most prominent explanations of the changing political climate – the so-called rise of ‘fake news’ and biased information. I use this explanation to show how my analytical framework can help distinguish between different political dimensions. By examining it within my framework and conceptualising the empirical evidence, I argue that the change in political behaviour is not located within the locutionary dimension.

The misinformation explanation

One of the most common explanations given for the growing amount of polarisation within the public has been the spreading phenomenon of ‘fake news’ (Allcott and Gentzkow, 2017; Hang Au et al., 2022; Ruffo et al., 2023; Spohr, 2017). While the introduction of social media and the decentralisation of information outlets has offered unprecedented access to information, it has also become increasingly difficult to reliably ensure that consumed information is derived from credible and trustworthy sources. Therefore, while citizens now have access to large amounts of information, they also are subject to being influenced by an unprecedented amount of disinformation. Proponents of misinformation explanations usually understand people’s disagreement arising from different preferences or information friction (Azzimonti and Fernandes, 2023). This theoretical view of political disagreement shares the interpretive framework and often explicitly uses models from what I have labelled the ‘issue-based’ group of theories. Here, political disagreement between two people usually originates in their different preferences. However, it can also form through information differences, where people with the same preferences come to different conclusions due to consuming different media and, therefore, possessing different and maybe misleading or false information. When viewing disagreement within this framework, a behaviour change, while preferences stay constant, must necessarily be due to a change in information (Allcott and Gentzkow, 2017; Azzimonti and Fernandes, 2023). The changing media setting offers a convincing cause for this change as it often seems that political disagreement has reached the basis of our democratic foundation – shared factual beliefs. This seems evident in much of the data that proponents of these hypotheses put forward. For example, when Americans were asked factual statements in a political context, such as ‘Has American GDP increased or decreased under the previous president?’, respondents reflected a clear partisan bias to answer in favour of their political party (Hannon, 2021). This means that if the previous president was a Democrat, those responding that GDP had decreased were far more likely to be Republicans, whereas those saying it has increased were more likely to be Democrats.

Analysing this within my theoretical framework, this seems like a case of locutionary disagreement. Two groups disagree about a specific, measurable issue and, based on this, form their political opinions. As this is a quantifiable issue, this form of disagreement should consist of a simple information problem. In contrast to complex normative positions, factual disagreement should be easily resolvable by providing convincing evidence to either side. However, voters are relatively invulnerable to changing their beliefs, even when provided with conclusive data to the contrary (Achen and Bartels, 2016). However, when offered a financial incentive to give correct answers in response to the same factual questions, almost all partisan bias vanished, and even partisan respondents gave correct responses to the same set of factual questions (Hannon, 2021). If, when given an incentive to do so, respondents provide accurate answers to questions they previously answered falsely, their intention cannot have been the assertion of truth. And consequently, their disagreement cannot have been within the locutionary dimension. It consisted of something other than an information problem. Therefore, disagreement must have originated from a different dimension – the illocutionary one. Rather than seeing it as locutionary disagreement, we should shift our perception to seeing the illocutionary speech act behind the responses. It then becomes more than a simple question about the economy but rather a question of which party they support. Instead of taking the answer ‘Obama has ruined our economy’ to be a factual assertion about the state of the economy, it might instead contain an illocutionary speech act expressing ‘Go Republicans!’.

Data from the American National Election Studies (ANES) and other experimental surveys tracking the issue beliefs of the US electorate seem to support the hypothesis that the change in political behaviour is not occurring within the locutionary dimension. Data from the ANES, which tracks voters’ self-reported positions on numerous central political issues, shows little change in polarisation on issues such as the size of government, federal spending on social security, increasing taxation for millionaires, views on gun control policy, US military involvement abroad or health care (ANES, 2024). In total, these datasets suggest that the US electorate is no more extreme in its political views than they were 50 years ago (Fiorina et al., 2008). On the other hand, identification with political parties has rapidly increased over the last 20 years, with less than 50 per cent identifying as either liberal or conservative between the start of the dataset in 1972 and 2004, compared to a rapid increase to more than 60 per cent during the 2020 presidential election (ANES, 2024). Simultaneously, data shows that, on average, Americans perceive the difference in issue positions to be around 200 per cent higher than they actually are (Levendusky and Malhotra, 2016). Although there has been some debate around the extent to which issue polarisation is occurring (Abramowitz and Saunders, 1998), the data offers a convincing account that issue polarisation has stayed rather constant (Dias and Lelkes, 2022; Fiorina et al., 2008; Iyengar et al., 2012) and that there is little evidence suggesting issue polarisation is occurring in the general public outside of the political elites (Hannon, 2021). Instead, it is the feeling towards the rival political party that has drastically changed. Whereas the view of their own party has remained rather constant, voters view the other party as increasingly negative. Data from the ANES measuring the warmth voters feel out of 100 points towards their own and the opposing party shows that, while sentiments towards their own party have remained constant at around 71 points, feelings towards the rival party have dropped from 48 points in 1974 down to 19 points in 2020 (ANES, 2024). Compared to ‘issue polarisation’, this has been called ‘affective polarisation’ (Garzia and Ferreira da Silva, 2022), where the public is not polarised on a concrete issue but behaves in a polarised way.

Principally, affective polarisation could be driven by either locutionary or illocutionary factors. One could, for example, be hostile towards another person because one feels strongly about an issue or because of the strong connection one feels towards a political party – similar to how one can be influenced by one’s emotional attachment to a sports team. Interpreting these results within the theoretical framework, we can distinguish the dimensions influencing the changing perlocutionary behaviours, such as an increase in hostile attitudes towards political out-groups. Considering that issue positions remained constant, these changes cannot be ascribed to factors from the locutionary dimension. Irrespective of how much absolute influence the locutionary dimension has on political behaviour, the findings that locutionary factors remain constant while the perlocutionary dimension is changing suggest that change is driven by factors from the illocutionary dimension. The changing perlocutions are not influenced by bottom-up locutionary influences but rather by top-down illocutionary changes.

Partisan sorting

When conceptualising the changing political environment within the analytical framework, it becomes clear that change must originate within the illocutionary dimension. As set out in the section discussing the theoretical models of voting behaviour above, those theories that emphasise illocutionary factors influencing political behaviour primarily identify social identity and group connections as the primary influences (Achen and Bartels, 2016; Campbell, 1960; Lazarsfeld, 1944). Indeed, empirical data has found a strong link between the degree to which a voter’s political party is a part of their social identity and their degree of affective polarisation (Mason, 2013; Kalmoe and Mason, 2022). Consider a soldier who might identify strongly with the military, which may lead her to feel more strongly about issues connected to military spending or military engagement abroad.

Now, as social identity has been argued to be the most significant channel within the defined illocutionary dimension, changes in this dimension are likely linked to changes in social identities or their relation to politics. One avenue that has been explored within the empirical literature is that of voter alignment. Contrary to the soldier example above, most individuals will not simply be part of one social identity but instead identify as part of a plethora of different social groups and identities to varying extents, in which political parties only make up one part. For example, think of a working-class individual who is also Christian. She might be sympathetic towards the Democratic Party because of her working-class background and her affinity to labour unions, but also towards the Republican Party because of the importance she places on her faith and the traditional values she associates with. This is an example of a ‘cross-cutting social identity’ (Mason, 2016), an individual who identifies with groups and identities that cross the party line. These individuals act as moderating factors in their parties, decreasing the social distance between their political party and social groups typically associated with the opposing party (Mason, 2015; Röllicke, 2023). Furthermore, they increase the distance between identities within a single political party, as more social groups associated with a party will lead to less alignment between the individual’s own social identities and those within their party. More diversity within a political party, therefore, simultaneously increases the distance between a voter and their own party while decreasing the distance between their party and the opposing one.

In recent years, however, political parties have become increasingly socially homogenous, and the distance between them in terms of the social identities they represent is growing (Lang and Pearson-Merkowitz, 2015). The differences between the ethnic identities of Democrats and Republicans have increased from a 26 per cent difference in 1980 to 36 per cent in 2012, and religious differences between the parties have also doubled in the same period (Mason, 2016). These two mark some of US society’s most influential social identities (Achen and Bartels, 2016). At the same time, social differences within political parties are rapidly decreasing (Harteveld, 2021; Levendusky, 2009). As argued above, these factors lead to stronger feelings of identification between party members and the political party while also leading to greater distance to the opposing party. Therefore, socially and ideologically sorted partisans express far stronger political sentiments and hostility to political out-groups than unsorted partisans (Ciuk, 2023; Mason, 2016).

Within my framework, this can be conceptualised as follows. We start with the status quo, where the locutionary and illocutionary dimensions influence perlocutionary acts to different degrees, which, for simplicity, we assume are constant. Now, through partisan sorting, the personal meaning that we connect with politics increases, represented by a strengthening of the influence from the illocutionary dimension. Relative to before, this means that the top-down influences – the influences from the illocutionary on the perlocutionary dimension – have strengthened. Consequently, our current political actions should be understood as illocutionary expressions to a greater extent than before.

Conclusion

This paper introduced a novel analytical framework that distinguishes between the locutionary, perlocutionary and illocutionary of dimensions of political speech acts. By applying this framework to the debate on political polarisation and the growing political hostility, the paper provided a more nuanced understanding of the origins and nature of these phenomena. Specifically, it challenged the prevalent focus on the locutionary dimension – issue-based disagreements – and instead argues that the growth in hostility is primarily driven by changes in the illocutionary dimension. While the paper suggests partisan sorting as a key driver of these illocutionary changes, future research might explore other factors within the illocutionary dimension that might contribute to this trend. For instance, the role of media consumption and the impact of social media use could be analysed with a focus on their illocutionary effects. By expanding this line of research, we might be able to gain a more accurate understanding of the factors driving political hostility.

Glossary

American National Election Studies (ANES): A collection of US national surveys conducted before and after every Presidential election dating back to 1948. The consistency of questions, high number of participants and long timespan make the ANES the gold standard within political science for tracking US public opinion over time.

Austin’s speech act theory: A theory from the philosophy of language in which Austin argued that there are three distinct acts present whenever we speak. The locutionary act, containing the literal content of our words, the perlocutionary act, containing the consequences of our words, and the illocutionary act, containing the meaning of our words. He describes these as three different speech acts that can be analysed and conceptualised as distinct from each other.

Disinformation: This refers to the deliberate spread of false or misleading information with the intent to deceive or manipulate public opinion.

Downsian model: Developed by economist Anthony Downs, this refers to a model of political competition where political parties and voters are rational actors seeking to maximise votes. It is based on the idea that in a two-party system, parties will adopt policies that appeal to the ‘median voter’ – the individual whose preferences lie at the centre of the political spectrum – in order to win the election. The model emphasises voter self-interest and economic motivations in voting choice.

Illocutionary act: Originally used to describe the speech act containing the meaning behind of our words, within this article it is used to describe the political speech acts containing political meaning, such as implicit or explicit considerations about social identities.

Locutionary act: Originally used to describe the speech act containing the content of our words, within this article it is used to describe the political speech acts referring to political content, such as the holding of certain issue beliefs or beliefs surrounding a party’s competence at handling an issue.

Perlocutionary act: Originally used to describe the speech act containing the consequences of our words, within this article it is used to describe the political speech acts containing expressive political behaviour, such as engaging in debates or voting.

Political speech act: This refers to any statement or declaration made by a political actor that not only conveys information but also performs an action, such as making a promise, issuing a demand or declaring support. It draws from speech act theory, where language is seen as an action that can influence reality.

To cite this paper please use the following details: Knoth, D. (2024), 'The Dimensions of Political Speech – Conceptualising the Origins of Political Hostility', Reinvention: an International Journal of Undergraduate Research, Volume 17, Issue 2,https://reinventionjournal.org/index.php/reinvention/article/view/1456/1363. Date accessed [insert date]. If you cite this article or use it in any teaching or other related activities please let us know by e-mailing us at Reinventionjournal@warwick.ac.uk.