“I Have the Best Words”:
Ideological Differences as Expressed in Politicians’ Language
Anna Brosius Student number: 10841652
a.brosius@uva.nl
24.06.2016
Master Thesis
Research Master Communication Science University of Amsterdam
Abstract
Political conservatism is generally associated with a higher need for closure, which is reflected in language use. Specifically, conservatives prefer language structures that offer more stability and closure, namely increased use of nouns and decreased use of complexity. Extant evidence focused exclusively on one-dimensional political systems, with the one structuring ideological line ranging from liberal (left-wing) to conservative (right-wing). This does not account for the complexity of modern European party systems. In particular, it overlooks pro- vs. anti-European ideology as an important dimension of political conflict that is not related to the liberal–conservative dimension. The present study takes the
multi-dimensional nature of European political system into account and theorizes that language use correlates with pro- vs. anti-European ideology. Using a Python script, 845 speeches,
delivered by politicians from seven German parties between 2006 and 2015, are automatically analyzed for the level of complexity and the share of nouns. The results show that a higher use of nouns in politicians’ speeches is indeed correlated with anti-European ideology. The use of linguistic complexity however is not related to pro- or anti-European ideology. Contrary to the expectations, politicians of conservative parties use more complex language. The use of nouns and complexity does not differ over the years, challenging the hypothesis that language use could change during situations that cause insecurity, specifically the
economic crisis. The findings contribute to understanding the appeal of anti-European parties, ideological differences expressed in language, and the ideological conceptualization of multi-dimensional European party spaces.
“I Have the Best Words”:
Ideological Differences as Expressed in Politicians’ Language
Euroscepticism is one of the biggest challenges for the European Union in the early 21st century. In the wake of the ongoing crisis of the EU, almost all member states have seen the rise of Eurosceptic parties, for example the French Front National, the Dutch Partij voor de Vrijheid, the British Ukip, the Hungarian Jobbik, or the Danish Dansk Folkeparti (Stokes, 2015). As of mid-2016, some countries – such as Great Britain or Greece – are even
discussing an exit from the EU or the Eurozone.
But why do Eurosceptic parties apparently have such great appeal to voters? As Mutz, Sniderman, and Richard (2006, p. 1) contended, “politics, at its core, is about persuasion”. In order to understand the roots of their rise, it is worthwhile to investigate how parties
propagating an anti-European ideology communicate vis-à-vis the electorate. Politicians are more likely to persuade when they cater to their voters’ “psychological needs” (Cichocka, Bilewicz, Jost, Marrouch, & Witkowska, 2016, p. 6), which are based on their personality. Voters’ needs and personality traits differ according to their political ideology (B. N. Bakker & de Vreese, 2016; Caprara & Zimbardo, 2004; Schoen, 2007; Schoen & Schumann, 2007). In particular, conservative citizens (Caprara & Zimbardo, 2004; Jost et al., 2007; Jost, Glaser, Kruglanski, & Sulloway, 2003) as well as conservative politicians (Caprara, Barbaranelli, & Zimbardo, 2002; Caprara & Zimbardo, 2004) show a higher need for closure than liberals. In other words, they show a greater preference for stability and structure as opposed to liberals’ greater tolerance of ambiguity (Webster & Kruglanski, 1994).
How, then, might Eurosceptic politicians cater to the psychological needs of their voters? A pivotal part of political communication is the language that politicians use.
“Language is virtually all a politician has to wield influence with.” (Lowndes, 2016) Evidence suggests that conservative politicians can voluntarily or involuntarily satisfy their voters’ increased need for closure by using a linguistic style that conveys stability and structure, such
as a higher share of nouns (Cichocka et al., 2016) and lower linguistic complexity (Mandel, Axelrod, & Lehman, 1993). Indeed, conservative politicians’ language in public speeches is characterized by a greater usage of nouns (Cichocka et al., 2016) and lower complexity (Cichocka et al., 2016; Tetlock, 1983, 1984) than liberal politicians’.
However, evidence on language differences between politicians have exclusively distinguished between liberal (left-wing) and conservative (right-wing) politicians (Cichocka et al., 2016; Tetlock, 1984, 1984). This classic distinction rests on the assumption that
political ideology is a one-dimensional construct. Yet in many European countries, there is considerable ideological tension not only along the classical left–right dimension, but also along the dimension of being pro- vs. anti-European (van der Brug & van Spanje, 2009). Whether there are differences in language use along this second dimension is a question that extant research has left unanswered.
This question represents the starting point of the present thesis. Taking into account the multidimensional nature of the European party system, it investigates whether pro vs. anti- European ideology is related to differences in politicians’ language use. The guiding
hypothesis is that anti-European politicians will display more linguistic structures that provide structure and stability than pro-European politicians. It is tested against the competing
hypothesis put forth by the extant literature – that the main ideological difference in language use is that between conservative and liberal politicians.
The core of the investigation consists of an analysis of speeches given by politicians from parties that differ both on the conservative–liberal dimension and the pro- vs. anti-European dimension. By investigating language differences across both dimensions, this study expands the theory of how ideology is expressed in linguistic structures, taking into account the multi-dimensional nature of European political systems. The results can enable a deeper understanding of the underlying psychological mechanisms of the communication of
Eurosceptic politicians and whether their appeal to voters might be related to their language use satisfying their voters’ psychological needs.
Theoretical Background
The Interplay of Personality and Ideology
Ideology is often conceptualized as a one-dimensional construct, ranging from
“conservative” to “liberal” (Jost, 2009), or from “left” to “right”. Conservatism, as opposed to liberalism, consist of two mains aspects: First, opposition to change and second, a preference for hierarchy and increased acceptance of inequality (Jost et al., 2003). These preferences, particularly opposition to change, seem to be rooted within certain personality traits. Previous research shows that personality and ideology – or similar concepts – are related. For example, personality traits of voters can change their opinions about foreign and security policies (Schoen, 2007) or the EU (B. N. Bakker & de Vreese, 2016) and increase their support for or identification with parties (B. N. Bakker, Hopmann, & Persson, 2015; Schoen & Schumann, 2007). Ideology and personality do not only coincide among voters but also among political elites, as evidenced by the finding that politicians display different personality patterns depending on their ideology (Best, 2011; Dietrich, Lasley, Mondak, Remmel, & Turner, 2012).
One personality trait that is particularly related to conservatism is need for closure, sometimes also referred to as need for structure, stability, or security. It refers to an
individual’s preference for order, structure, and predictability (Webster & Kruglanski, 1994) and often results in attempts to avoid or reduce uncertainty and ambiguity (Jost et al., 2003). For individuals with a high need for closure, any structuring information is preferable over ambiguity (Kruglanski, Webster, & Klem, 1993). Therefore, high need for closure is related to more simple cognitive structures that ignore ambiguity (Hiel & Mervielde, 2003).
Conservatives often have an increased need for security and stability, which renders them more skeptical of change than liberals (Jost, Federico, & Napier, 2009; Jost et al., 2003). They also have a stronger preference for stability and an increased adversity to threat and uncertainty (Jost, 2009; Jost et al., 2003). Taking many complex categories into account can spark ambiguity and uncertainty; therefore, conservatism is associated with less ambiguous thoughts (Eidelman, Crandall, Goodman, & Blanchar, 2012). In conclusion, conservatives have a higher need for closure than liberals.
Personality as Reflected in Language Use
Language reflects personality; the way humans speak gives away part of who they are (Tausczik & Pennebaker, 2010). Simple words and grammatical structures, such as pronouns or prepositions (Pennebaker, Mehl, & Niederhoffer, 2003; Weintraub, 1989), but also the use of certain words, for example emotional words or self-references (Fast & Funder, 2008), are associated with a variety of personality types or characteristics. Just like other personality traits, need for closure is reflected in language. More specifically, a higher need for closure entails (1) lower integrative complexity of language and (2) more frequent use of nouns.
Integrative complexity, i.e., the complexity of an individual utterance, partly depends on the context of the utterance; yet, integrative complexity also reflects the speaker’s
cognitive complexity (Suedfeld, 2010). Cognitive complexity is a relatively stable trait, with higher levels of complexity denoting information processing with a “combination of
flexibility, high levels of information search, and tolerance for ambiguity, uncertainty, and lack of closure” (Suedfeld, 2010, p. 1670). Integrative complexity thus reflects the extent to which different and potentially conflicting viewpoints are taken into account when crafting an argument (Abe, 2011; Brundidge, Reid, Choi, & Muddiman, 2014; Suedfeld & Leighton, 2002; Tausczik & Pennebaker, 2010). As laid out before, individuals with a high need for closure are less likely to tolerate competing and potentially contradicting arguments, prefer
simple cognitive structures and are therefore also less likely to use a high level of integrative complexity when speaking.
Nouns categorize objects into relatively stable and unidimensional groups, whereas adjectives emphasize properties of an object, which usually has more than one property and can change (Carnaghi et al., 2008; Gelman & Heyman, 1999; Hall & Moore, 1997; Walton & Banaji, 2004). Furthermore, nouns do not allow for varying degrees of a characteristic or relative statements. For example, using adjectives, a person can be described as conservative, more conservative or the most conservative. This is not possible using nouns, which stand for less flexible and dichotomous categories – a person is either a conservative or not a
conservative. Using nouns to describe members of a group makes group membership more salient and accentuates differences compared to out-group members (Graf, Bilewicz, Finell, & Geschke, 2013). The use of nouns therefore prevents complex cognitive structures that take multiple aspects into account (Carnaghi et al., 2008). This linguistic evidence suggests that individuals with higher need for closure have a preference for nouns, as they stand for stability and simple structures.
Language Differences Related to Political Ideology
Generally, people with different political ideologies use different linguistic cues; for example, conservatives use less emotional words than liberals (Sylwester & Purver, 2015). Given that personality traits are reflected in language, and that conservatives differ from liberals on particularly in need for closure, conservatives and liberals should differ in their language regarding language complexity and the use of nouns.
There is some evidence for such liberal–conservative differences in language use: On the elite level, conservative politicians tend to make less complex statements than liberals (Tetlock, 1983, 1984) and political bloggers use less complex language than their liberal counterparts (Brundidge et al., 2014). Also in the general population, conservative citizens
use less integratively complex language than liberals do (Mandel et al., 1993), possibly because of a higher need for closure (Jost et al., 2003; Sidanius, 1985). Regarding the use of nouns, a first study by Cichocka et al. (2016) showed that conservative citizens prefer nouns in speech and conservative politicians use more nouns in their speeches, as nouns convey stability and unambiguity, which satisfies conservatives’ higher need for closure.
Complexity of European Party Spaces
Extant studies on language differences between politicians of different ideologies heavily rely on the one-dimensional conceptualization of ideology, ranging from
conservatism to liberalism, or right and left (Jost, 2009). This broad distinction emerged in the US-American context. Although it captures major differences between political parties, it does not take the multi-dimensional nature of European political systems into account. This multi-dimensional nature entails that there is competition along multiple lines of ideology and that there are more parties occupying different niches in the political space. Is this ideological complexity also reflected in language differences between politicians? In other words: Are different ideological stances in a multidimensional political system predicting language differences, analogous to the way the liberal-conservative spectrum may predict language use in one-dimensional systems?
At the mass level, i.e. for ordinary citizens, it is more common among scholars to conceptualize political ideology as a multidimensional construct. Most commonly, they distinguish between a social or cultural and an economic dimension (Carney, Jost, Gosling, & Potter, 2008; Feldman & Johnston, 2014; Johnston & Wronski, 2015; Kriesi et al., 2006; Malka, Soto, Inzlicht, & Lelkes, 2014). At the elite level, however, i.e. for highly politically sophisticated individuals, politicians, and parties these two dimensions tend to correlate strongly (Converse, 2006; Jost et al., 2009; van der Brug & van Spanje, 2009), forming a one-dimensional construct. The anchors of this one-one-dimensional construct of political ideology are
usually referred to as left and right or liberal and conservative. For political parties, the Chapel Hill Expert Survey (CHES) indicates that the correlations of the social and economic dimensions with the general liberal–conservative dimension are r = .79 and r = .78
respectively (R. Bakker et al., 2015). In other words, these two ideological dimensions do not differ considerably from the general liberal–conservative assessment of parties, which means that one dimension is sufficient to reflect different party stances towards social and economic issues.
European ideological conflicts are, however, structured along a second dimension. Specifically, van der Brug and van Spanje (2009) proposed that, aside from the liberal– conservative dimension, party positions are structured by their stance on European
integration. For instance, Figure 1 visualizes data from the CHES, which show that German party stances on European integration are not strongly correlated (r = .44) with their positions on the liberal–conservative dimension (R. Bakker et al., 2015). On the liberal part of the spectrum (ranging from 0 to 5 on the liberal–conservative scale), there are pro-European parties – such as the SPD and Die Grünen – but also the anti-European party Die Linke. The same pattern emerges on the conservative side (ranging from 5 to 10 on the liberal–
conservative scale): There are pro-European forces, namely CDU, CSU, and FDP, as well as anti-European parties, such as the AfD and NPD.
Figure 1. German party positions.
Author’s illustration, based on data from CHES 2014 (R. Bakket et al., 2015).
The presence of a second dimension of political conflict in European politics poses an important tension in the expectations about the language use of politicians in the multi-dimensional party space. If only conservatives have a higher need for closure and the second structuring component – being pro or anti-European – is not relevant, then language
differences that are based on higher need for closure should only exist between liberals and conservatives, for example SPD and CDU politicians. This pattern was previously found in one-dimensional political system regarding the use of nouns (Cichocka et al., 2016) and complexity (Cichocka et al., 2016; Tetlock, 1983, 1984). It leads to the following first hypothesis.
H1: Conservative politicians use (a) lower levels of integrative complexity and (b) more nouns in their language than liberal politicians.
However, there is a second dimension – anti-European ideology – that structures the European party space. This second ideological dimension could also be related to need for
closure and therefore be expressed in the language use of politicians in a similar way. In other words, differing party positions on issues of European integration may have similar correlates (e.g. need for closure and linguistic differences) as party positions on the liberal–conservative dimension in terms of personality traits. Hence, the question arises whether being pro- or anti-European is related to differential language use. The following section will discuss evidence that the pro- vs. anti-European dimension of ideology might be an important predictor for language use of politicians.
In Germany, all anti-European parties (NPD, AfD and Die Linke) have one vital aspect in common: They all aim to provide psychological shelter and security from threats that could stem from an open Europe. While right-wing parties appeal to those high in need for closure because of their conservative cultural views (Kossowska & Hiel, 2003; Malka et al., 2014), left-wing parties appeal to those with high need for closure because of their economic views “aimed at providing material security and protection” (Malka et al., 2014, p. 1044). This makes anti-European parties attractive for individuals that want to avoid the insecurities of an expanding Europe and strive for stability and structure. In other words, anti-European parties’ politicians and their voters might share a higher need for closure, similar to conservative parties’ politicians and voters. Voluntarily or involuntarily, anti-European politicians might cater to this increased need for closure among their voters by using language that appeals to those with a high need for closure.
A second characteristic these anti-European parties share is an at least moderate degree of populism (Hough & Koß, 2009; Rooduijn & Pauwels, 2011; Zaslove, 2008).
Populist leaders are distinct from other political leaders because of their simpler language use, which sets them apart from the more complex language of establishment politicians and brings them closer to the common citizens (Bos & Brants, 2014; Bos, van der Brug, & de Vreese, 2013). As Zaslove (2008, p. 327) put forward, “[populist leaders] are proud to speak in an uncomplicated language and in the language of the tavern, the café and the street. This is
contrasted to the over-sophisticated and the overly ideological language of the more traditional political leaders.” The mentioned studies did not explicitly investigate pro- and European parties, but rather populist parties. As there is a considerable overlap of European and populist parties in Germany, the findings nevertheless suggest that anti-European parties might use less complex language.
In sum, the stance toward Europe is a second dimension that structures the European party space. An anti-European ideology could be related to a higher need for closure which, in turn, might be reflected in the language use of politicians from anti-European parties.
Compared to pro-European politicians, anti-European politicians might show a preference for linguistic structures that provide stability and security: a decreased use of integrative
complexity and an increased use of nouns. Traditionally, scholars have only focused on language differences between liberals and conservatives (Cichocka et al., 2016; Tetlock, 1983, 1984). Yet, the introduced evidence leads to the following competing hypothesis, which takes the multi-dimensional nature of European party systems into account.
H2: Anti-European politicians use (a) lower levels of integrative complexity and (b) more nouns in their language than pro-European politicians.
Need for Stability in Situations of Crisis
In the previous sections, it was established that conservatism and higher need for closure are expressed in language. However, need for closure as well as the associated conservatism can change over time to an extent. For example, economically disadvantaged individuals – for whom an economic crisis is most threatening – show increases in
conservative ideology in times of recession, possibly due to stronger feelings of insecurity, which in turn increase the situational need for closure (Milojev, Greaves, Osborne, & Sibley, 2015). This increased need for closure among citizens might in turn be expressed in the language of politicians, who are catering their voters’ needs.
There is indeed some evidence that political actors use simpler language in unstable times. For example, former chairman of the US Federal Reserve Alan Greenspan showed lowered language complexity in his statements as the economic situation deteriorated (Abe, 2011). Similarly, the integrative complexity in statements of Tony Blair and George W. Bush decreased after the 9/11 terrorist attacks (Suedfeld & Leighton, 2002). New York mayor Rudolph Giuliani used simpler language in times of crisis (Pennebaker & Lay, 2002).
Furthermore, integrative complexity of politicians’ speeches decreases in the months prior to the outbreak of a war (Suedfeld, 2010; Suedfeld & Leighton, 2002).
To summarize, this evidence suggests that a crisis, which causes insecurity and therefore also need for closure, might in turn lead politicians to use less complex language and more nouns. There are several possible reasons for such changes in language use during economic crises. On one hand, politicians may feel more insecure themselves and therefore be more prone to using language that promises stability. On the other hand, they may
deliberately or intuitively choose to use less complex language as a rhetorical strategy in order to appeal to voters’ heightened need for stability and closure, portraying themselves as
reliable leaders (Abe, 2011; Suedfeld, 2010). Taken together, this evidence leads to the following hypothesis.
H3: In times of heightened economic and social insecurity, politicians will use (a) lower integrative complexity and (b) more nouns in their language.
Method
Case Selection
In order to test differences in the language of politicians in a multi-dimensional European party system, a sample of speeches from German politicians was created. The German context is an apt case for the present research endeavor for both conceptual and practical reasons. First, rising anti-European sentiments are largely unrelated to the traditional
left–right dimension in Germany, as there are both anti-European right-wing (AfD) and left-wing (Die Linke) parties (see Figure 1). Thus, the distinction between anti- and pro-European ideology is a clearly discernible second ideological dimension in the German context. This enables the research project to contribute to closing the gap in academic literature – whether this second ideological dimension is also reflected in the language of politicians. Second, Germany experienced a rise of anti-European sentiments following the economic crisis, especially with the prospect of another bailout of Greece in 2010, which made the crisis highly salient (BBC, 2015; Finanzen, 2015). Being one of the biggest countries in Europe and one of the most powerful political forces, Germany also is an important pillar of European integration. Strong anti-European sentiments in Germany thus pose a threat to the European project, rendering it a task of high societal relevance to understand such sentiments. Finally, there is a practical reason for the choice: In Germany, parliamentary speeches are publicly accessible.
Sample of Speeches
The speeches were mostly sampled from the national parliament, the Bundestag. All members of the Bundestag participate in its plenary sessions (Deutscher Bundestag, 2016c). They are protocoled verbatim and the protocols are publicly available online. One session – one parliamentary day – contains approximately 50 to 60 speeches; the speaking time of all parties is proportional to the number of seats that the party has in parliament (Deutscher Bundestag, 2016a). The order of the speeches in the protocols are according to the order in which they were given; usually rotating between politicians from different parties. Table 1 displays the distribution of seats in the German Bundestag during the three relevant legislative periods, namely 2005 – 2009, 2009 – 2013 and 2013 onwards (Deutscher Bundestag, 2016b). It shows that the CDU/CSU and the SPD held most seats, whereas Die Linke, Die Grünen and the FDP only held a fraction of the seats.
Table 1
Distribution of seats in the German Bundestag between 2005 and 2016 Legislative
period Overall Die Linke Die Grünen SPD CDU/CSU FDP
2005 – 2009 614 54 51 222 226 61
2009 – 2013 622 76 68 146 239 93
2013 – 631 64 63 193 311 0
Three years were chosen for the sample to reflect different states of the European financial crisis: 2006 as a relatively calm pre-crisis year, 2010 as a year in which the Euro crisis climaxed with Greece’s near bankruptcy and 2015 as a post-crisis year in which the Euro crisis was not a salient issue anymore (Finanzen, 2015). 2010 should therefore be the year that caused most insecurity, whereas 2006 should be the year that caused least insecurity and accordingly need for closure. For each year, five plenary protocols are randomly selected, using a random number generator.
Because of the minimum electoral threshold of 5%, not all parties that are relevant to the German political landscape are present in the Bundestag. In order to represent one right-wing, anti-European party, the sample also includes speeches from the party Alternative für Deutschland (AfD), which is only represented in the parliaments of some federal states but not in the Bundestag. 20 speeches from five randomly selected protocols were sampled from the parliament of the state Saxony and 26 speeches from four randomly selected protocols were sampled from the parliament of the state Thuringia, in which the AfD has 14 (of 126) and eight (of 91) representatives respectively. Because the party only became a relevant political force recently (Gathmann & Wittrock, 2016), AfD speeches are only included for the year 2015. Even though the regional parliaments are different from the Bundestag, the
regional parliaments in Thuringia and Saxony include some of the most prominent AfD politicians, such as Bernd Höcke, Frauke Petry, and André Barth, representing the party elite. Therefore, the composition of the AfD in these federal states is not strongly different from the
composition of parties in the Bundestag, which include both party leaders and less prominent members.
A total of 845 speeches was collected, with 124 from Die Linke, 124 from Die Grünen, 200 from the SPD, 180 from the CDU, 66 from the CSU, 105 from the FDP and 46 from the AfD. On average, the speeches have 61.64 sentences (SD = 30.06, Min = 7,
Max = 254), 891,78 words (SD = 412,14, Min = 100, Max = 3603) and 5281,27 letters (SD = 2367.20, Min = 568, Max = 20896) letters. 249 speeches are from 2015, 357 from 2010, and 240 from 2006. The unequal group size of the years is due to the varying number of speeches in the randomly selected protocols.
One concern that can be raised when studying politicians’ speeches is that many politicians do not write their speeches themselves. However, one can assume that
speechwriters are part of the “inner circle” – the closest team around the politician – and share similar ideological views and personality traits (Cichocka et al., 2016). Therefore, it is
possible to draw inferences about political elites that are grouped into different ideologies through their party membership.
Data Collection and Coding
After downloading the protocols from the online archives, all speeches were manually copied from the protocols into a dataset. During the collection process, all direct quotes consisting of at least one full sentence within the speeches were excluded, as they do not reflect the speaker’s own language. Based on the protocols, different variables were manually coded. This includes the date of the debate, the name of the politician who gave the speech, the party she or he belongs to, and the topic of the debate. This factual information is included in each protocol; therefore, intercoder reliability is not a question of concern for this form of coding, as the information is simply copied from the protocols. Errors that might have
occurred during the process of data collection are random and will therefore not bias the results in a systematic manner.
The indicators of language complexity and the use of nouns were analyzed automatically. There are several measures for language complexity, such as the Flesch Kincaid score, which mostly focus on readability metrics like sentence and word length (Kincaid, Fishburne, & Chissom, 1975). They usually do not include elements that account for the complexity behind a statement, for example by accounting for words that indicate competing arguments or viewpoints within one statement. However, there is one measure that accounts for this vital aspect of integrative complexity: The Linguistic Inquiry and Word Count Program (Abe, 2011). Following Abe (2011), the degree of language complexity is an index consisting of three elements: (1) the number of words per sentence, (2) the number of letters per word, and (3) the number of words per sentence that indicate a complex thought behind a statement because they imply competing ideas. In English, these words include expressions such as “on one hand”, “but”, “also”, “although”, “except” (Abe, 2011; Tausczik & Pennebaker, 2010). For the purpose of the present study, this list was translated to German, including possible equivalents and alternative expressions with the same or similar meanings. All German words included in the analysis can be found in Appendix A. The final complexity score is the mean of the number of words per sentence, the number of letters per word and the number of indicator words per sentence, with all three variables being z-standardized across the full sample, in order to give them the same weight in the index.
The convergent validity of this complexity measure was tested by calculating the correlation with another established measure of language complexity: The Flesch Kincaid grade (Kincaid et al., 1975). The analysis was performed using the textstat package in Python. The correlation between the Flesch Kincaid grade and the complexity measure used here is r = .84. Therefore, convergent validity is given. This correlation is visualized in Figure 2. Moreover, the Flesch-Kincaid grade can facilitate the interpretation of the level of complexity
used in the political speeches: The Flesch Kincaid grade stands for the years of schooling that are necessary to understand a given text without difficulty. The average Flesch Kincaid grade of all speeches is 10.95 (SD = 1.88, Max = 18.60, Min = 6.40). This means that the average speech is rather complex and only easily understandable for well-educated individuals. A complexity score of –1 corresponds to a Flesch Kincaid grade of approximately 7; a score of 0 corresponds to approximately 11, and 1 corresponds to approximately 14.
Figure 2. Correlation between complexity scores and Flesch Kincaid grade.
A common way to search for nouns and other parts of language is to use a so-called Part-of-Speech (POS) tagger (e.g., Cichocka et al., 2016). The problem with POS taggers, however, is that they are usually only available in English and not open access. This also implies that it is a black-box approach; it is unclear how nouns are identified, and the taggers are potentially incomplete. For these reasons – language barriers, access and completeness – the present study relied on an alternative approach.
In German, all nouns are capitalized. There are no other word groups that are capitalized, with the exception of abbreviations and the first word in a sentence. In order to avoid counting every letter of an abbreviation as one noun, capital letters were only counted if they are directly followed by a lowercase letter and not by another capital letters. The solution for the problem concerning the beginning of sentences was more complex. A typical sentence in German does not start with a noun. Not only is this uncommon, but the nouns will usually also require an article in front. But grammatically, there is the possibility that a sentence starts with the plural form of a noun or an indefinite noun, which does not always require an article. However, as this is not likely to distort the results as much as the low-quality POS taggers, it is an acceptable degree of imprecision to exclude all capitalized words in the beginning of sentences from the analysis, assuming that only a small minority of them will actually be nouns. An additional benefit of this method of searching for nouns is that it is very straightforward and does not rely on any black-box approach from a third party.
In order to compute the variable indicating the share of nouns in a speech, the number of sentences was deducted from the total number of capital letters, which gives a good indication of the number of nouns in the speech. The number of nouns was then divided by the total number of words in the next, which resulted in the share of nouns of the given speech.
The validity of the automated noun detection was tested as follows: One random speech was selected (ID 398) and all nouns were manually counted by three human coders. One of the coders was the author, the only instruction that the other two coders received was to count all nouns in the text. All coders are native speakers of German and university-level educated. The automated analysis lead to a noun count of 171, whereas the two human coders counted 170 (author), 167 (coder 1), and 167 (coder 2) nouns. As these numbers are fairly close, the measure could be deemed valid.
In summary, a self-written Python code1 (see Appendix B) counted the number of capital letters, the number of letters overall, the number of words, the number of sentences, and the frequency of the indicators words for integrative complexity for every speech in the sample. Because the automated analysis of language complexity and nouns in the sampled speeches did not include any subjective elements, classical intercoder reliability parameters are not applicable.
Results
Descriptive Statistics
Before conducting the statistical analysis, the properties of the two dependent
variables “complexity” and “nouns” are assessed. The dependent variable “complexity” is an index of three z-standardized subcomponents with a mean of zero (SD = .71), ranging
from -1.64 to 3.47. Speeches that are less complex than average therefore have a negative complexity score, whereas more complex speeches have a positive complexity score. As visualized in Figure 3, the distribution is slightly skewed to the right (Skew = .78) and slightly peaked (Kurtosis = 1.09). There are more outliers with very high language complexity than outliers with very low language complexity.
The variable “nouns” indicates the share of nouns in a speech, i.e. what percentage of all words used are nouns. On average, speeches have a share of nouns of M = .23 (SD = .03), ranging from .16 to .33. As visualized in Figure 3, the distribution is negligibly skewed to the right (Skew = .36) and negligibly peaked (Kurtosis =.15). There is a small but statistically significant correlation between the two dependent variables “nouns” and “complexity” (r(845) = .29, p < .01). Substantially, this means that a more complex speech also contains more nouns per sentence. Furthermore, the small correlation coefficient indicates that the two variables measure different constructs.
Figure 3. Distributions of the dependent variables “Nouns” and “Complexity”. Differences in Language Use Between Parties and Across Years
In order to test mean-level differences in complexity and nouns between the parties as well as between the different time points, a multivariate analysis of variances (MANOVA), relying on the general linear model, was performed in the statistical program SPSS. First, general assumptions of the ANOVA were tested; substantive results are present in the subsequent paragraph. The Levene’s test of equality of error variances showed that equal variances can be assumed for both dependent variables, “nouns” (F(17, 827) = 1.18, p = .28) and “complexity” (F(17, 827) = 1.54, p = .08). The omnibus F-test confirms that the model is significant for both the dependent variable “complexity” (F(17) = 2.22, p < .01) as well as the dependent variable “nouns” (F(17) = 3.58, p < .01). Wilk’s multivariate significance tests indicate that the independent variables “party” (λ(12) = .93, p < .01) and “year” (λ(4) = .96, p < .01) are statistically significant predictors of both complexity and nouns. However, the interaction between party and year is not a statistically significant predictor of any of the two dependent variables (λ(18) = .98, p = .50).
The substantive results – the between-subject effects – can be found in Table 2. These results show that language complexity differs significantly by party, although the effect is rather small, based on the statistical coefficient Eta-squared. There are no significant
differences in language complexity between the three years under study. As concerns use of nouns, there are small significant differences both between the parties as well as between the years.
Table 2
Party Differences in Language Use (Between-subjects effects) Dependent variable Independent variable df F p Partial Eta-squared Complexity Intercept 1 .85 .358 .00 Party 6 4.12 .000 .03 Year 2 2.17 .115 .01 Party × Year 9 1.15 .323 .01 Nouns Intercept 1 55755.04 .000 .99 Party 6 6.48 .000 .05 Year 2 13.49 .000 .03 Party × Year 9 .66 .746 .01
In order to investigate these relationships in more depth, post-hoc tests are performed. For the pairwise comparisons of the parties, the least significant difference (LSD) test was used, as this test does not require equal group sizes. The LSD test is a rather liberal post-hoc test. However, seeing that the number of comparisons is relatively high due to the seven different parties, adjusted tests – such as the Bonferroni test – are likely to be too conservative and to increase the probability of Type II errors disproportionally (Garson, 2015). In order to perform pairwise comparisons of the years, the LSD test is used as well for consistency. However, because the independent variable “year” only comprises three groups, the results of the LSD test are validated using the Bonferroni post-hoc test in this case.
Figure 4 visualizes differences in language complexity between the parties. Speeches delivered by politicians of Die Linke exhibit the lowest language complexity, being
significantly less complex than that of speeches by politicians of SPD (p = .039), CDU (p = .001), CSU (p = .002), and AfD (p < .001). They are, however, not significantly less complex than speeches by politicians of Die Grünen (p = .104) and FDP (p = .157). AfD politicians use the most complex language, being more complex than politicians of all other parties (Grüne: p = .006, SPD: p = .007, FDP: p = .006), with the exception of CDU
(p = .064) and CSU (p = .310). Other pairwise comparisons do not lead to any further statistically significant party differences in language complexity. All pairwise comparisons can be found in Appendix C.
These results neither confirm H1a, which stated that conservative politicians use less complex language than liberal politicians, nor the competing hypothesis H2a, which stated that anti-European politicians use less complex language than pro-European politicians. Both hypotheses are thus rejected. There is no relation between pro- or anti-European ideology and language complexity. Furthermore, regarding the liberal–conservative spectrum of ideology, the results show the contrary of what was hypothesized: Conservative politicians use more complex language than liberal politicians do. Table 3 shows that the effect of the party on language complexity is rather small when considering the statistical coefficient (η2 = .03). However, the mean difference between Die Linke and the AfD is .47, which equals 61.04% of a standard deviation and therefore a considerable share of the total variance of the measure.
Figure 4. Complexity scores per party. Complexity scores are z-standardized.
Next, the differences in use of nouns are examined. Figure 5 shows that politicians of Die Linke and AfD use the most nouns in their speeches. The difference between these two parties is statistically non-significant (p = .755). Furthermore, Die Linke politicians are not significantly different from the CSU (p = .062), but use significantly more nouns than all other parties (Grüne: p = .011, SPD: p < .001, CDU: p = .047, FDP: p = .002). AfD politicians use significantly more nouns than politicians from other parties (Grüne: p = .028, SPD:
p = .001, FDP: p = .010), with the exception of CDU (p = .083) and CSU (p = .078). All further pairwise comparisons can be found in Appendix D; none of them are statistically significant, with the exception of the difference between the CDU and SPD (p = .011).
These results do not support Hypothesis 1b, which stated that conservative politicians use more nouns than liberal politicians, but they do support the competing Hypothesis 2b, which stated that anti-European politicians use more nouns than pro-European politicians. The two parties that strongly oppose European integration – Die Linke and AfD – use
significantly more nouns than parties that are decisively pro-European, such as Die Grünen, SPD, and FDP (see Figure 1).
Again, the effect of party on the use of nouns is rather small when considering the statistical coefficient (η2 = .05). However, again, one should consider the rather small variance of linguistic structures. The standard deviation of use of nouns is .027. The mean difference between, for example, AfD politicians and SPD politicians is .014, which equals 51,85% and therefore a considerable share of the total variance of the measure.
Figure 5. Share of nouns per party.
The range of the y-axis was adjusted for better visibility.
Finally, the main analysis showed that there is no difference in language complexity between the three years under study. Therefore, Hypothesis 3a, which claimed that politicians use lower levels of complexity in times of insecurity, is rejected. However, there is a
difference in use of nouns over the years. Figure 6 illustrates the average use of nouns in the three different years. The LSD post-hoc test shows that politicians used significantly fewer nouns in the year 2015 than they did in the years 2006 and 2010. There are no significant
differences in the use of nouns between the years 2010 and 2006. According to the Bonferroni post-hoc test, only the years 2010 and 2015 differ significantly from each other. As there are only three groups, the LSD test might not be conservative enough and the significant
difference between 2015 and 2006 should only be considered a tentative result. Therefore, the pattern is rather unclear. At any rate, the results do not support Hypothesis 3b, which claimed that politicians use more nouns in times of insecurity, i.e. the year 2010.
Figure 6. Share of nouns per year.
The range of the y-axis was adjusted for better visibility. Discussion
This study asked whether political ideology manifests in language of politicians due to underlying personality differences in need for closure. Focusing on the sample case of
Germany, it tested the hypotheses that (a) conservative politicians, or (b) anti-European politicians, exhibit a stronger need for structure, stability and closure in their speeches than liberal or pro-European politicians, as indicated by a lower linguistic complexity and a stronger reliance on nouns.
The results yielded only mixed support for these hypotheses. As concerns the use of nouns, the results showed that anti-European politicians do in fact use more nouns than pro-European politicians, whereas conservative politicians do no differ from liberal politicians in their use of nouns. In other words, pro- vs. anti-European ideology is related to the use of more nouns, which indicate higher need for closure. However, there was a different pattern for the second indicator, language complexity. Contrary to previous findings (Cichocka et al., 2016; Tetlock, 1983, 1984), conservative politicians used higher, not lower, levels of
complexity than liberal politicians. There was no relationship between use of linguistic complexity and pro- vs. anti-European ideology. These results suggest that the fault line between pro-vs. anti-European ideology in Germany is only linked to the use of nouns, but not to linguistic complexity; and that linguistic complexity is related to left–right ideology differently than expected. The results will be interpreted in more depth in the following paragraphs.
Interpretation of Results and Implications
Use of nouns. The finding that politicians from anti-European parties, both left-wing and right-wing, used most nouns lends further support to the idea that political ideology manifests in language use. It thereby extends previous research by Cichocka et al. (2016), who found that conservative politicians use more nouns than liberal politicians in the one-dimensional political context of the US, to the multi-one-dimensional political contexts of Europe. The explanation for these findings on the basis of previous research is that anti-European parties provide their voters with psychological shelter from outside threats, regardless of their general liberal-conservative ideology (Malka et al., 2014). This is expressed in language, particularly through an increased use of nouns, which convey stability and closure (Carnaghi et al., 2008; Cichocka et al., 2016). Their language use caters to the increased need for closure
of their voters, who are more inclined to vote for politicians who satisfy this need (Caprara & Zimbardo, 2004).
The results also imply that the pro- vs. anti-European dimension of ideology in European party spaces, as proposed by van der Brug and van Spanje (2009), is indeed an important structuring element and, as hypothesized in the present study, related to the use of nouns in politicians’ language. This suggests that in a European, multi-party context, a one-dimensional conceptualization of ideology, i.e., liberal vs. conservative (Converse, 2006; Jost et al., 2009), is not sufficient to describe ideological differences and their manifestation in language.
Language complexity. Regarding language complexity, the present study’s results did not confirm either of the competing hypotheses, and they contradict previous findings on the language complexity of politicians of different ideologies. Other scholars found that liberal politicians (Cichocka et al., 2016; Tetlock, 1983, 1984) and bloggers (Brundidge et al., 2014) use more complex language than conservatives do. The speeches analyzed in the present study show the opposite pattern: More conservative parties use more complex language, which in turn indicates more complex ideas that are being expressed in their language (Abe, 2011; Tausczik & Pennebaker, 2010). Many scholars suggest that a higher need for closure is an important reason for avoiding complex thought (Eidelman et al., 2012; Jost, 2009; Jost et al., 2003). Does this suggest that conservative politicians in Germany have a lower need for closure than liberal politicians? This conclusion may seem unwarranted given previous
evidence about the higher need for closure of conservatives (Best, 2011; Jost, 2009; Jost et al., 2003).
An alternative interpretation is that other factors shape the level of language
complexity of German politicians more than their need for closure does. There are two main potential confounding factors. First, the level of education of politicians and their voters might play a role. Assuming that some components of linguistic complexity – such as word
length and sentence length – increase as the level of education increases (Kincaid et al., 1975), education could be an important predictor of language complexity. However, there is very little variance in the education of German members of parliament. The vast majority of deputies, namely 90,5%, have university degrees (Jauch & Sturm, 2015). This is in line with the generally high complexity of their speeches. But do politicians adapt their linguistic complexity to the level of education of their audience at least to some extent? Extant evidence on voters’ educational levels does not fit with this hypothesis. Voters of Die Grünen are the most educated (43,5% have a high school degree), followed by the FDP (36,1%), Die Linke (35,5%), the AfD (21,2%), the SPD (18,6%), the CDU/CSU (16,4%), and extreme right-wing parties at the bottom of the table (Brähler & Decker, 2014). The pattern is relatively stable over the years (Brähler & Kruse, 2011). There is no recognizable relationship between politicians’ language complexity and the voters’ educational levels; therefore, the factor education is ruled out.
A second factor impacting language complexity could lie in the nature of the speeches. While previous studies used speeches (Cichocka et al., 2016; Tetlock, 1983) that were clearly targeted at a broader audience, the present study relied on parliamentary speeches. Even though the speeches are also directed at the public, the most salient audience is the parliament itself, comprising politicians from different ideologies. It might be that politicians are less inclined to use language that appeals to their voters’ needs in such a context and more inclined to use language that may persuade politicians from other parties. This might only affect language complexity (and not the share of nouns), because language complexity is something that is more likely to be consciously adapted to the audience than the share of nouns.
Changes in language use over time. The study did not find convincing support for the hypothesis that integrative language complexity and nouns change systematically over time, i.e., in situations that cause insecurity. A potential reason for this is that the language
differences are reflections of stable personality traits. In some previous studies, language use was found to be relatively stable intra-individually (Fast & Funder, 2008; Mehl &
Pennebaker, 2003). Therefore, the results of this study could lend support to the hypothesis that language differences mostly reflect stable personality traits, such as need for closure, that are not very prone to change over time. Alternatively, because individuals with high
socioeconomic status, such as politicians, do not personally feel more insecure during a crisis (Milojev et al., 2015), another explanation could be that in this case, the increased need for closure of the subgroup of disadvantaged voters might not be enough of a stimulus for politicians to adapt their general language use.
Limitations
Future research should address several limitations of the present study. One limitation concerns the comparability of the speeches, in regard to which two issues bear mentioning: First, the AfD speeches were collected in a different context than all other speeches, namely two regional parliaments. As mentioned before, these two regional representations consist of some of the top members of the party, making the party compositions similar to the smaller parties in the Bundestag. However, the regional parliaments at least partly discuss different topics than the Bundestag, which might affect the choice of linguistic structures. This issue might be solvable in the future, in case the AfD manages to become part of the Bundestag. The second comparability issue concerns all parties: The composition of big parties like the CDU and SPD might be different than the composition of smaller parties like the FDP or CSU. Due to the sheer number of delegates, the sample of the bigger parties are likely more diverse. Individual outliers have less of an impact on the average language use of the the bigger parties. Therefore, it would be interesting to extend the present study with results about the language use of exclusively party, who might be more comparable across parties.
It is also important to note that the results allow conclusions about political elites. As such, they do not allow for inferences about the general population. Politicians are different from the average citizen, as a result of self-selection and voter selection. For example, evidence on the personality traits of German politicians suggests that members of parliament are more open and and extroverted, but less agreeable and conscientious than the general population (Best, 2011). Furthermore, as demonstrated by the results, the language use itself is not comparable; the politicians generally used highly complex language. Future research could shed further light on this issue by comparing the language use of politicians and their voters. This could provide insight into whether there are also differences in the language use of ordinary citizens along the lines of political ideology.
Conclusion
What explains the recent rise in support for Eurosceptic parties in Europe? Based on the notion that persuasion is an essential part of politics (Mutz et al., 2006), scholars from different disciplines have investigated multiple aspects of political persuasion, such as the voters’ political beliefs and schemata (Lau, Smith, & Fiske, 1991), the quality of political arguments (Cobb & Kuklinski, 1997), and the tone of messages (Arceneaux & Nickerson, 2010). With regard to the persuasive success of anti-European parties in particular, scholars have identified several contributing factors, such as the lack of a European identity (Weßels, 2007), low political trust, feelings of threat (Lubbers & Scheepers, 2007), but also context factors, such as a country’s economic situation (Serricchio, Tsakatika, & Quaglia, 2013).
Communication scientist studying political persuasion mostly focus on media
selection (Arceneaux, Johnson, & Cryderman, 2013) and media effects (de Vreese, 2007; Hill et al., 2013; Zaller, 2006). The present study shows that psychological characteristics do not only shape media selection and responses to media content (Valkenburg & Peter, 2013); but that media content – specifically, political speeches – can also indicate psychological
characteristics of the speaker. These mediated characteristics might influence the persuasion of recipients. Therefore, not only the psychological traits of the recipient, but also those of the sender of a message play a role in the persuasion process.
Based on this perspective, the present study points out another potential factor driving Eurosceptic politicians’ success. Through an increased use of nouns – which convey stability and structure – these politicians may cater to their voters’ “psychological needs” (Cichocka et al., 2016, p. 6), satisfying the higher need for closure of anti-European voters and providing them with psychological shelter from outside threats. Understanding this dynamic can also help pro-European parties in winning voters by better tailoring their language – their most powerful tool of persuasion (Lowndes, 2016) – to their voters’ needs. They might benefit from providing alternative forms of security and shelter or from emphasizing the benefits of a united Europe for voters with a higher need for closure.
References
Abe, J. A. A. (2011). Changes in Alan Greenspan’s language use across the economic cycle: A text analysis of his testimonies and speeches. Journal of Language and Social Psychology, 30(2), 212–223.
Arceneaux, K., Johnson, M., & Cryderman, J. (2013). Communication, persuasion, and the conditioning value of selective exposure: Like minds may unite and divide but they mostly tune out. Political Communication, 30(2), 213–231.
Arceneaux, K., & Nickerson, D. W. (2010). Comparing negative and positive campaign messages: Evidence from two field experiments. American Politics Research, 38(1), 54–83.
Bakker, B. N., & de Vreese, C. H. (2016). Personality and European Union attitudes: Relationships across European Union attitude dimensions. European Union Politics, 17(1), 25–45.
Bakker, B. N., Hopmann, D. N., & Persson, M. (2015). Personality traits and party
identification over time. European Journal of Political Research, 54(2), 197–215. Bakker, R., Edwards, E., Hooghe, L., Jolly, S., Marks, G., Polk, J., … Vachudova, M. (2015).
2014 Chapel Hill Expert Survey. Chapel Hill, NC: University of North Carolina. Retrieved from chesdata.eu
BBC. (2015, November 8). Greece profile - Timeline. BBC. Retrieved from http://www.bbc.com/news/world-europe-17373216
Best, H. (2011). Does personality matter in politics? Personality factors as determinants of parliamentary recruitment and policy preferences. Comparative Sociology, 10(6), 928–948.
Bos, L., & Brants, K. (2014). Populist rhetoric in politics and media: A longitudinal study of the Netherlands. European Journal of Communication, 29(6), 703–719.
Bos, L., van der Brug, W., & de Vreese, C. H. (2013). An experimental test of the impact of style and rhetoric on the perception of right-wing populist and mainstream party leaders. Acta Politica, 48(2), 192–208.
Brähler, E., & Decker, O. (2014). Die Parteien und das Wählerherz 2014. Universität Leipzig. Retrieved from http://home.uni-leipzig.de/decker/waehlerherz_2014_grafiken.pdf Brähler, E., & Kruse, J. (2011). Die Parteien und ihre Anhänger. Universität Leipzig.
Retrieved from http://www.uni-leipzig.de/pdf/pm2011-343_ergebnisse.pdf Brundidge, J., Reid, S. A., Choi, S., & Muddiman, A. (2014). The “Deliberative Digital
Divide”: Opinion leadership and integrative complexity in the U.S. political blogosphere. Political Psychology, 35(6), 741–755.
Caprara, G. V., Barbaranelli, C., & Zimbardo, P. G. (2002). When parsimony subdues distinctiveness: Simplified public perceptions of politicians’ personality. Political Psychology, 23(1), 77–95.
Caprara, G. V., & Zimbardo, P. G. (2004). Personalizing politics: A congruency model of political preference. American Psychologist, 59(7), 581–594.
Carnaghi, A., Maass, A., Gresta, S., Bianchi, M., Cadinu, M., & Arcuri, L. (2008). Nomina sunt omina: On the inductive potential of nouns and adjectives in person perception. Journal of Personality and Social Psychology, 94(5), 839–859.
Carney, D. R., Jost, J. T., Gosling, S. D., & Potter, J. (2008). The secret lives of liberals and conservatives: Personality profiles, interaction styles, and the things they leave behind. Political Psychology, 29(6), 807–840.
Cichocka, A., Bilewicz, M., Jost, J. T., Marrouch, N., & Witkowska, M. (2016). On the grammar of politics - or why conservatives prefer nouns. Political Psychology. doi:10.1111/pops.12327
Cobb, M. D., & Kuklinski, J. H. (1997). Changing minds: Political arguments and political persuasion. American Journal of Political Science, 41(1), 88–121.
Converse, P. E. (2006). The nature of belief systems in mass publics (1964). Critical Review, 18(1-3), 1–74.
Deutscher Bundestag. (2016a). Geschäftsordnung. Retrieved March 21, 2016, from https://www.bundestag.de/bundestag/aufgaben/rechtsgrundlagen/go_btg
Deutscher Bundestag. (2016b). Members of the 18th German Bundestag. Retrieved March 21, 2016, from http://www.bundestag.de/htdocs_e/bundestag/members18
Deutscher Bundestag. (2016c). Plenary business in the German Bundestag. Retrieved March 21, 2016, from http://www.bundestag.de/htdocs_e/bundestag/plenary
de Vreese, C. H. (2007). A spiral of Euroscepticism: The media’s fault? Acta Politica, 42(2-3), 271–286.
Dietrich, B. J., Lasley, S., Mondak, J. J., Remmel, M. L., & Turner, J. (2012). Personality and legislative politics: The Big Five trait dimensions among U.S. State Legislators. Political Psychology, 33(2), 195–210.
Eidelman, S., Crandall, C. S., Goodman, J. A., & Blanchar, J. C. (2012). Low-effort thought promotes political conservatism. Personality and Social Psychology Bulletin, 38(6), 808–820.
Fast, L. A., & Funder, D. C. (2008). Personality as manifest in word use: Correlations with self-report, acquaintance report, and behavior. Journal of Personality and Social Psychology, 94(2), 334–346.
Feldman, S., & Johnston, C. (2014). Understanding the determinants of political ideology: Implications of structural complexity. Political Psychology, 35(3), 337–358. Finanzen. (2015). Aktueller GfK-Konsumklimaindex. Retrieved May 23, 2016, from
http://www.finanzen.net/gfk/
Garson, G. D. (2015). GLM Multivariate, Manova & Canonical Correlation. Asheboro, NC: Statistical Associates Publishing.
Gathmann, F., & Wittrock, P. (2016, March 13). AfD triumphiert, schwarzer Sonntag für die CDU. Spiegel Online. Retrieved from
http://www.spiegel.de/politik/deutschland/landtagswahlen-2016-in-der-blitzanalyse-afd-triumph-cdu-debakel-a-1082090.html
Gelman, S. A., & Heyman, G. D. (1999). Carrot-eaters and creature-believers: The effects of lexicalization on children’s inferences about social categories. Psychological Science, 10(6), 489–493.
Graf, S., Bilewicz, M., Finell, E., & Geschke, D. (2013). Nouns cut slices: Effects of linguistic forms on intergroup bias. Journal of Language and Social Psychology, 32(1), 62–83.
Hall, D. G., & Moore, C. E. (1997). Red bluebirds and black greenflies: Preschoolers’ understanding of the semantics of adjectives and count nouns. Journal of Experimental Child Psychology, 67(2), 236–267.
Hiel, A. van, & Mervielde, I. (2003). The need for closure and the spontaneous use of complex and simple cognitive structures. The Journal of Social Psychology, 143(5), 559–568.
Hill, S. J., Lo, J., Vavreck, L., & Zaller, J. (2013). How quickly we forget: The duration of persuasion effects from mass communication. Political Communication, 30(4), 521– 547.
Hough, D., & Koß, M. (2009). Populism personified or reinvigorated reformers? The German Left Party in 2009 and beyond. German Politics & Society, 27(2), 76–91.
Jauch, M., & Sturm, D. F. (2015, April 28). Ohne Studium in die hohe Politik? Fast unmöglich. Die Welt. Retrieved from
http://www.welt.de/politik/deutschland/article140168531/Ohne-Studium-in-die-hohe-Politik-Fast-unmoeglich.html
Johnston, C. D., & Wronski, J. (2015). Personality dispositions and political preferences across hard and easy issues. Political Psychology, 36(1), 35–53.
Jost, J. T. (2009). “Elective affinities”: On the psychological bases of left–right differences. Psychological Inquiry, 20(2-3), 129–141.
Jost, J. T., Federico, C. M., & Napier, J. L. (2009). Political ideology: Its structure, functions, and elective affinities. Annual Review of Psychology, 60(1), 307–337.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychological Bulletin, 129(3), 339–375.
Jost, J. T., Napier, J. L., Thorisdottir, H., Gosling, S. D., Palfai, T. P., & Ostafin, B. (2007). Are needs to manage uncertainty and threat associated with political conservatism or ideological extremity? Personality and Social Psychology Bulletin, 33(7), 989–1007. Kincaid, J. P., Fishburne, R. P., & Chissom, B. S. (1975). Derivation of new readability
formulas (Automated Readability Index, Fog Count, and Flesch Reading Ease Formula) for Navy enlisted personnel. In Research Branch report 8-75. Memphis: Naval Air Station.
Kossowska, M., & Hiel, A. van. (2003). The relationship between need for closure and conservative beliefs in Western and Eastern Europe. Political Psychology, 24(3), 501– 518.
Kriesi, H., Grande, E., Lachat, R., Dolezal, M., Bornschier, S., & Frey, T. (2006).
Globalization and the transformation of the national political space: Six European countries compared. European Journal of Political Research, 45(6), 921–956.
Kruglanski, A. W., Webster, D. M., & Klem, A. (1993). Motivated resistance and openness to persuasion in the presence or absence of prior information. Journal of Personality and Social Psychology, 65(5), 861–876.
Lau, R. R., Smith, R. A., & Fiske, S. T. (1991). Political beliefs, policy interpretations, and political persuasion. The Journal of Politics, 53(3), 644–675.
Lowndes, N. (2016, June 18). Double-plus effective. Why Donald Trump’s rhetoric — with apologies to Orwell — works so well. The Economist. Retrieved from
http://www.economist.com/news/books-and-arts/21700610-why-donald-trumps-
rhetoricwith-apologies-orwellworks-so-well-double-plus-effective?fsrc=scn%2Ffb%2Fte%2Fpe%2Fed%2Fdoublepluseffective
Lubbers, M., & Scheepers, P. (2007). Explanations of political Euroscepticism at the individual, regional and national levels. European Societies, 9(4), 643–669.
Malka, A., Soto, C. J., Inzlicht, M., & Lelkes, Y. (2014). Do needs for security and certainty predict cultural and economic conservatism? A cross-national analysis. Journal of Personality and Social Psychology, 106(6), 1031–1051.
Mandel, D. R., Axelrod, L. J., & Lehman, D. R. (1993). Integrative complexity in reasoning about the Persian Gulf War and the accountability to skeptical audience hypothesis. Journal of Social Issues, 49(4), 201–215.
Mehl, M. R., & Pennebaker, J. W. (2003). The sounds of social life: A psychometric analysis of students’ daily social environments and natural conversations. Journal of
Personality and Social Psychology, 84(4), 857–870.
Milojev, P., Greaves, L., Osborne, D., & Sibley, C. G. (2015). Stability and change in political conservatism following the global financial crisis. Personality and Social Psychology Bulletin, 41(1), 127–139.
Mutz, D. C., Sniderman, P. M., & Richard, A. B. (Eds.). (2006). Political persuasion and attitude change. Ann Arbor: Univ. of Michigan Press.
Pennebaker, J. W., & Lay, T. C. (2002). Language use and personality during crises: Analyses of mayor Rudolph Giuliani’s press conferences. Journal of Research in Personality, 36(3), 271–282.
Pennebaker, J. W., Mehl, M. R., & Niederhoffer, K. G. (2003). Psychological aspects of natural language use: Our words, our selves. Annual Review of Psychology, 54(1), 547–577.
Rooduijn, M., & Pauwels, T. (2011). Measuring populism: Comparing two methods of content analysis. West European Politics, 34(6), 1272–1283.
Schoen, H. (2007). Personality traits and foreign policy attitudes in German public opinion. Journal of Conflict Resolution, 51(3), 408–430.
Schoen, H., & Schumann, S. (2007). Personality traits, partisan attitudes, and voting behavior. Evidence from Germany. Political Psychology, 28(4), 471–498.
Serricchio, F., Tsakatika, M., & Quaglia, L. (2013). Euroscepticism and the global financial crisis. JCMS: Journal of Common Market Studies, 51(1), 51–64.
Sidanius, J. (1985). Cognitive functioning and sociopolitical ideology revisited. Political Psychology, 6(4), 637–661.
Stokes, B. (2015, June 2). The rise of nontraditional, Eurosceptic parties. Retrieved June 14, 2016, from
http://www.pewglobal.org/2015/06/02/chapter-4-the-rise-of-nontraditional-eurosceptic-parties/
Suedfeld, P. (2010). The cognitive processing of politics and politicians: Archival studies of conceptual and integrative complexity. Journal of Personality, 78(6), 1669–1702. Suedfeld, P., & Leighton, D. C. (2002). Early communications in the war against terrorism:
An integrative complexity analysis. Political Psychology, 23(3), 585–599.
Sylwester, K., & Purver, M. (2015). Twitter language use reflects psychological differences between Democrats and Republicans. PLOS ONE, 10(9), e0137422.
Tausczik, Y. R., & Pennebaker, J. W. (2010). The psychological meaning of words: LIWC and computerized text analysis methods. Journal of Language and Social Psychology, 29(1), 24–54.
Tetlock, P. E. (1983). Cognitive style and political ideology. Journal of Personality and Social Psychology, 45(1), 118–126.
Tetlock, P. E. (1984). Cognitive style and political belief system in the British House of Commons. Journal of Personality and Social Psychology, 46(2), 365–375. Valkenburg, P. M., & Peter, J. (2013). The Differential Susceptibility to Media Effects
Model. Journal of Communication, 63(2), 221–243.
van der Brug, W., & van Spanje, J. (2009). Immigration, Europe and the “new” cultural dimension. European Journal of Political Research, 48(3), 309–334.
Walton, G. M., & Banaji, M. R. (2004). Being what you say: The effect of essentialist linguistic labels on preferences. Social Cognition, 22(2), 193–213.
Webster, D. M., & Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. Journal of Personality and Social Psychology, 67(6), 1049–1062.
Weintraub, W. (1989). Verbal behavior in everyday life. New York: Springer.
Weßels, B. (2007). Discontent and European identity: Three types of Euroscepticism. Acta Politica, 42(2), 287–306.
Zaller, J. (2006). The myth of massive media impact revived: New support for a discredited idea. In D. C. Mutz, P. M. Sniderman, & A. B. Richard (Eds.), Political persuasion and attitude change (pp. 17–79). Ann Arbor: Univ. of Michigan Press.
Zaslove, A. (2008). Here to stay? Populism as a new party type. European Review, 16(3), 319–336.
