Spyros Kosmidis is a Post-doctoral Research Fellow in Political Science and Computational Linguistics at the Department and Politics and International Relations, University of Oxford. He is currently working on the project: “The Radicalisation of Greek Politics: Violence, Extremism and Support for Democracy”. Here he outlines a different research project on the topic of exploring the affective norms of political party manifestoes:
This is a research project that aspires to measure and understand the affective norms of political discourse. We already know that political parties propose policies to persuade the electorate. In fact, all else being equal, the party developing the most persuasive set of policies will win the election. But, we also know that persuasion is not only about numbers, policy proposals and goals. Rhetoric has an affective aspect that is difficult to measure and understand. How do political parties present their policies to make them more appealing and persuasive? And, how can political scientists extract these pieces of information? With respect to the first question, we argue that on top of the programmatic content of political texts, political parties use emotional appeals to convince that their policies are superior. To measure that we examine how affective political rhetoric is through a content analysis of political language.
Our approach is to make use of what psychologists have done already in analysing emotions, in particular a psycholinguistic dictionary called ANEW (Affective Norms of English Words), which provides a way of interpreting and analysing the emotional content of words on a number of dimensions. We are currently interested in two emotional dimensions – Valence (which measures the direction) and Arousal (which measures the intensity). Examples of high valence would be words like triumphant, love or paradise, and low valence would be words like cancer, rejection or suicide. Examples of high arousal words would be rage or thrill, and low arousal would be found in words like fatigued or lazy. In a matter of minutes our software produces the aggregate affect scores to be analysed, so experimentation with alternative emotional dimensions (positive vs negative or enthusiasm vs anxiety) is an obvious and achievable aim for us. For now, we intend to apply the ANEW dictionary to political texts and speeches to extract over time measures of affective rhetoric, and in particular we will start with party manifestos – a very conservative test indeed – which are available in the UK going back to 1900 for the three main parties.
At the same time, we all may have an intuition that the role of emotions in party politics has grown in recent years, as parties often are less differentiated from one another on policy grounds and as the traditional anchors of class and party allegiance have declined. But little is really known about how parties use emotions, which ones ‘work’ electorally, and how emotional appeals have changed over time. Our preliminary results show that affective rhetoric increases when parties are indistinguishable in terms of policy positioning and decreases when the election outcome is not close. We also find that affective rhetoric is larger when more persuadable (undecided) voters exist in the electorate. Our analyses seem to be robust, and even though there are certain limitations, we are confident that our measures are not an artefact of language change over time or product of the alleged change in the way political communications operate. The mere fact that objective measures of the economy exert an intuitive –and robust—impact on affective rhetoric lends additional validity to our measure. These are important findings; they illuminate previously unrecognised patterns of party competition and shed light to the emotional aspect of politics that we know it exists, but we had been unable to understand.
Although an examination of how affective rhetoric or emotional appeals relate to politics could be easily implemented in a public opinion survey, the costs of such strategy would be enormous, let alone the fact that establishing long term patterns would be virtually impossible. On top of that, survey respondents (and human coders of political texts) tend to rely on their own dispositions to evaluate political speeches with respect to emotionality. This makes a variety of analyses susceptible to important biases. With this methodology, on the other hand, a wide range of research can be easily and effectively pursued.
For example, we plan to run some survey experiments to see how individuals react to different levels of affect (as measured with the ANEW dictionary). We also plan to reanalyse public opinion surveys (with a temporal component) to see how affective rhetoric influences individual voters’ actual political choices (rather than the whole electorate). Of course, we will extend our analyses to other English speaking democracies and see how systemic characteristics influence the relationships reported in the UK case. We will also move beyond party manifestos, -admittedly the least emotional collection of texts-, and analyse public debates, leadership speeches, media content, and public documents. We believe that by reporting how political actors present their policies will enhance our understanding of important democratic notions like representation and further our understanding of democratic politics.
Brader, Ted. 2006. Campaigning for hearts and minds: how emotional appeals in political ads work, Studies in communication, media, and public opinion. Chicago: University of Chicago Press.
Marcus, George E., W. Russell Neuman, and Michael MacKuen. 2000. Affective intelligence and political judgment. Chicago: University of Chicago Press.