Donald Trump speaks at the level of a fourth grader. So we were told in October 2015, when The Boston Globe announced the results of its study of nineteen presidential candidates’ announcement speeches. Having analyzed the speeches using the Flesch-Kincaid readability test, which determines the grade level required to understand a text, the newspaper found that Trump’s scored 4.1. Or, as the article gleefully proclaimed, “Trump’s speech could have been comprehended by a fourth-grader. Yes, a fourth-grader.” By contrast, Hillary Clinton’s announcement speech was rated three grade levels higher at 7.1, Marco Rubio’s at 8.6, Ted Cruz’s at 8.9, and Bernie Sanders’ at 10.1 (Viser Reference Viser2015).
“Right-wing populist uses simple language” headlines should come as no surprise. The dichotomy between a good “ordinary people” and bad “elites” is central to the concept of populism (Mudde Reference Mudde2007; Müller Reference Müller2016) and for many years, in what is now a vast literature, political scientists have affirmed as a self-evident truth that right-wing populists use simple language in order to appear closer to “ordinary people” and differentiate themselves from linguistically convoluted political elites (Canovan Reference Canovan1999; Taggart Reference Taggart, Mény and Surel2002; Bos, van der Brug, and de Vreese Reference Bos, van der Brug and de Vreese2011; Moffitt and Tormey Reference Moffitt and Tormey2014).Footnote 1 However, while the (rare) relevant empirical studies provide partial support for claims of greater right-wing populist linguistic simplicity, these focus on party manifestos rather than speeches by individuals (Bischof and Senninger Reference Bischof and Senninger2018) or are single-country case studies that for various reasons do not allow us to draw more general conclusions (Oliver and Rahn Reference Oliver and Rahn2016; Kayam Reference Kayam2018; Wang and Liu Reference Wang and Liu2018; Decadri and Boussalis Reference Decadri and Boussalis2019). Moreover, while there are important studies showing the greater simplicity of the arguments put forward by conservative politicians and activists, these neither tell us about the simplicity of the language nor focus specifically on populists (Tetlock Reference Tetlock1983, Reference Tetlock1984; Brundidge et al. Reference Brundidge, Reid, Choi and Muddiman2014). In short, the empirical research conducted to date casts some light on a topic otherwise characterized by received wisdom, but it cannot tell us whether greater linguistic simplicity is a common characteristic of right-wing populists.
Our study investigates precisely that issue. Specifically, we ask: Do right-wing populist leaders use simpler language than their principal mainstream competitors? To answer this, we assembled a corpus totaling approximately 1,000,000 words from speeches by ten recent right-wing populist and rival party leaders or presidential nominees from France, Italy, the United Kingdom, and the United States. This enabled us to go beyond previous studies and investigate more broadly whether right-wing populists do in fact use simpler language. For each leader in our four countries, we analyzed at least 100,000 words from their speeches over similar time periods, using the main measures employed by linguistics scholars for evaluating simplicity. Like other researchers, we deploy readability scales such as Flesch-Kincaid, but we go much further by also assessing lexical richness, lexical density and the presence of difficult words. By applying a comprehensive range of linguistic methods to a political science question, we thus seek to make a contribution to our understanding of populism.
The article proceeds as follows. In the next section, we set out the theoretical background regarding our conceptualization of right-wing populism and the long-standing claims about the simplicity of populist language. We then discuss some of the few studies that have looked, at least in part, at the simplicity of language used by populists compared to other parties and politicians. Following this, we introduce the data and methods we have used to analyze the linguistic simplicity of speeches by right-wing populist and other leaders: Donald Trump and Hillary Clinton in the United States; Marine Le Pen and Emmanuel Macron in France; Nigel Farage, David Cameron, and Ed Miliband in the United Kingdom; Matteo Salvini, Angelino Alfano and Matteo Renzi in Italy. In each case, our right-wing populists are the most significant exemplars from their respective countries over the past decade: President Trump, Le Pen (the Front National leader, who secured one-third of the vote in the 2017 presidential election), Farage (leader of the UK Independence Party, who was a key driver of Brexit), and Salvini (leader of the Northern League, which he has built into one of Italy’s major parties).
In our analysis section, we show how the assumptions and claims in the literature that right-wing populists use simpler language are not consistently borne out. While Trump’s language does tend to be simpler than Clinton’s on most measures, the gap is not so wide (for example, his campaign speeches are only one grade less on the Flesch-Kincaid scale than Clinton’s). In France and the UK, the results are more surprising. We find that Le Pen uses far more complex language than Macron according to all simplicity measures, while Farage is generally more complex than Miliband and Cameron. Finally, in Italy, we encounter a mixed picture. The right-wing populist Salvini is simpler than Renzi and Alfano according to readability scores, but he is not the simplest of the three on any of our other measures. We argue therefore that, while future studies of discourse or the complexity of ideas may reveal populists are indeed simpler in those senses than their rivals, the long-standing claim that populists are characterized by their use of simpler language than other politicians needs to be revised. In the concluding section, we discuss some of the implications of this for our understanding of populism and what further research in this area might usefully look at.
Right-Wing Populism and Simple Language
Why should we expect right-wing populists like Trump, Farage, Salvini, and Le Pen to use simpler language than their rivals? There are three main reasons. First, as noted earlier, the theoretical literature on populism claims that populists use simpler language than their mainstream “elite” rivals in order to appear closer to “the people”. Second, most of the (rare) empirical studies that have carried out linguistic analysis of right-wing populists compared to mainstream politicians suggest that they behave as the populism literature claims (Oliver and Rahn Reference Oliver and Rahn2016; Bischof and Senninger Reference Bischof and Senninger2018; Kayam Reference Kayam2018; Decadri and Boussalis Reference Decadri and Boussalis2019). Third, culturally conservative politicians and activists have been found to use simpler concepts (Tetlock Reference Tetlock1983, Reference Tetlock1984; Suedfeld et al. Reference Suedfeld, Bluck, Ballard and Baker-Brown1990) and simpler language than their liberal opponents (Schoonvelde et al. Reference Schoonvelde, Brosius, Schumacher and Bakker2019). Let us discuss these three reasons in turn.
While scholars disagree on whether populism is, inter alia, an ideology (Mudde Reference Mudde2007), a style (Moffitt and Tormey Reference Moffitt and Tormey2014), a discourse (Hawkins Reference Hawkins2009), or a strategy (Weyland Reference Weyland2001), there is reasonably broad agreement about its core features. As Canovan (Reference Canovan1981, 294) observes: “all forms of populism without exception involve some kind of exaltation of and appeal to ‘the People’ and all are in one sense or another anti-elitist.” This dichotomy of a “good” people and “bad” elites provides the basis for Mudde’s widely used definition of populism as “a thin-centred ideology that considers society to be ultimately separated into two homogeneous and antagonistic groups, ‘the pure people’ versus ‘the corrupt elite’, and which argues that politics should be an expression of the volonté générale (general will) of the people” (Mudde Reference Mudde2007, 23). While the “people vs. elites” is a fixed point for populism of both right and left, right-wing populism has a further antagonistic element: “the others.” Hence, for right-wing populists, the people are not only threatened from above by elites (political, financial, cultural, etc.) but also from below by a range of “others” whose identities, beliefs, or behaviors place them outside “the people” (Albertazzi and McDonnell Reference Albertazzi and McDonnell2015, 5-6).
Simple language is said to be at the heart of the juxtaposition between people and elites, with populists using it to distinguish themselves from elites and present themselves as being closer to the people. This is a consistently noted characteristic of populist politicians. Discussing research on populism, Rooduijn (Reference Rooduijn2014, 576-577) observes that for many other scholars
“the first feature of the allegedly populist style is the use of simplistic language. Because of their glorification of the people and their loathing of the elite, populists tend to use rather simple language that is understood by “normal” people and differs from the difficult and formal language of the elite.”
Populist linguistic simplicity has been a common assumption in the theoretical literature for at least two decades. For example, in one of the key articles within the field, Canovan (Reference Canovan1999, 5) noted populists’ use of “simple, direct language.” Populists do this, she argued, in order to contrast with establishment politicians’ “bureaucratic jargon” and to reflect the simplicity and directness of their own proposals. This is evidently not just a claim about the simplicity of ideas and arguments, but about the language used to convey them. Many other leading scholars of populism have made similar points. Taggart (Reference Taggart, Mény and Surel2002, 76) asserts that populists “strive for clarity, directness and simplicity,” while Moffitt and Tormey (Reference Moffitt and Tormey2014, 392) discuss populists’ “tendency towards simple and direct language” (see also Bos, van der Brug, and de Vreese Reference Bos, van der Brug and de Vreese2011, 187). There is also broad agreement on the reason why populists use simpler language than their mainstream rivals. Echoing Canovan, Zaslove (Reference Zaslove2008, 327) states that right-wing populists “speak in an uncomplicated language” to stand out from “the over-sophisticated and the overly ideological language of the more traditional political leaders” (see also Block and Negrine Reference Block and Negrine2017, 190). In short, as Bischof and Senninger (Reference Bischof and Senninger2018, 476) put it, the dominant political science view is that, for populists, “linguistic simplicity is a valuable tool for exhibiting the aloofness of the remaining political elite and fosters the impression of a strong bond between the populist and ordinary people.”
These assertions about populist linguistic simplicity tend to be presented as so intuitively obvious that they require no supporting evidence or further investigation. Nonetheless, in recent years, a handful of researchers have produced empirical work that sheds some light on the question. The first notable study of more than one country (but just one language) was that by Bischof and Senninger (Reference Bischof and Senninger2018). Analyzing party manifestos in Austria and Germany between 1945 and 2013 using a readability index, they find that, as expected, populist parties do “employ significantly less complex language” (Bischof and Senninger Reference Bischof and Senninger2018, 473). In particular, the linguistic simplicity of the Austrian Freedom Party’s manifestos increased substantially following its right-wing populist shift under new leader Jörg Haider in 1986 (Bischof and Senninger Reference Bischof and Senninger2018, 485). While helpful to our understanding of the simplicity of populist party communication over time, however, a comparison of manifestos cannot tell us whether populist leaders use simpler language than others (this, we should note, is not a criticism of the authors since examining spoken language was not their intention). Unlike speeches, manifestos are formal documents, which set out key positions and provide campaign discussion points for candidates and the media, but they are not intended for consumption by a wide audience of voters (Budge Reference Budge, Budge, Robertson and Hearl1987). Another single-language study is that by Decadri and Boussalis on members of the Italian Parliament. They find that, in the specific context of plenary parliamentary speeches (which have certain constraints of etiquette), “populist politicians, and members of populist parties, communicate using simpler language than their mainstream colleagues” (Decadri and Boussalis Reference Decadri and Boussalis2019, 16).
While there has been no other research to date on the comparative simplicity of right-wing populist language in Europe, there have been several studies that provide insights into the simplicity of Trump’s language. Although they focus more on questions such as populist attitudes among the public, Oliver and Rahn (Reference Oliver and Rahn2016, 194) also analyze the simplicity of language in the announcement of candidature speeches. They find that, compared to four other candidates in the Republican primaries and Hillary Clinton and Bernie Sanders, Trump generally used the simplest language (closely followed by Ben Carson and John Kasich). A second relevant recent study is that by Kayam (Reference Kayam2018), who, using a series of readability indices to examine transcripts of ten interviews and debates with Trump between July 2015 and March 2016, finds that Trump speaks at a level comprehensible to a fifth-grader, while five other candidates from the Republican and Democratic primaries do so at least at the level of a seventh-grader.
Wang and Liu, however, present a more mixed picture regarding Trump’s language. Having assembled a robust corpus for Trump and compared him to both Clinton and Barack Obama (the latter’s debates and speeches are taken from his 2012 re-election campaign), they find that “when debating with other candidates, Trump uses a smaller range of vocabulary, repeats his phrases and uses shorter words and shorter sentences” (Wang and Liu Reference Wang and Liu2018, 308). On the other hand, they discover that Flesch-Kincaid grade level results for his speeches are higher (9.32) than those for Clinton (8.04). They conclude, therefore, that “Trump may not be as undisciplined and crude as the common view suggests” (Wang and Liu Reference Wang and Liu2018, 309). Given that all but one of the fifty Trump speeches they analyze were delivered between March and October 2016, these results do not necessarily contradict those of Oliver and Rahn (Reference Oliver and Rahn2016) and Kayam (Reference Kayam2018), which were based on earlier periods (and different electoral contexts and text-types). Interestingly, Hawkins and Littvay (Reference Hawkins and Littvay2019, 15-17) also find that Trump’s speeches became more populist—through a greater focus on “people-centrism” rather than just “anti-elitism”—in the months after May 2016 (the period in which he secured the nomination and appointed Steve Bannon to his campaign team).Footnote 2
Finally, since right-wing populists are ideologically closer to conservatives than to liberals, it is worth noting that there is a long-standing body of literature showing the lesser complexity of concepts (known as “integrative complexity”) used in oral communication by conservative politicians compared to more liberal ones in a number of western democracies (Tetlock Reference Tetlock1983, Reference Tetlock1984; Suedfeld et al. Reference Suedfeld, Bluck, Ballard and Baker-Brown1990). For example, Tetlock (Reference Tetlock1984) found that, in the UK, “extreme conservatives” were less integratively complex than the three other ideological categories he identified: moderate conservatives, moderate socialists and extreme socialists. This may apply more widely to those on the right. Brundidge et al. (Reference Brundidge, Reid, Choi and Muddiman2014) have shown that conservative bloggers also use integratively simpler language in their posts than liberal ones do. In a more recent contribution to this literature, Schoonvelde et al. (Reference Schoonvelde, Brosius, Schumacher and Bakker2019) examined the relationship between ideology and linguistic complexity among liberals and conservatives. Using the Flesch-Kincaid scale to analyze speeches from a wide range of political leaders across Europe, they show that culturally conservative politicians tend to be less complex than culturally liberal ones. Since right-wing populists are also culturally conservative, Schoonvelde et al. (Reference Schoonvelde, Brosius, Schumacher and Bakker2019) thus provide more support for the claim that right-wing populists use simpler language than their opponents.
This brings us to a last point. What do we mean by “simple language”? In a nutshell, this means how easy it is to understand. Of course, simplicity is a relative concept: a text may be simple for some and difficult for others, depending on their familiarity with the contents it conveys and the lexical and syntactic features it deploys. Our approach uses a set of linguistic traits (including word and sentence length, lexical richness and density, and the presence of difficult words) as a proxy for measuring simplicity. This is in line not only with a long tradition in linguistics research, but also with the approach of those who have sought in recent years to analyze the simplicity of populist and other politicians’ language (Oliver and Rahn Reference Oliver and Rahn2016; Bischof and Senninger Reference Bischof and Senninger2018; Kayam Reference Kayam2018; Wang and Liu Reference Wang and Liu2018; Hawkins and Littvay Reference Hawkins and Littvay2019; Decadri and Boussalis Reference Decadri and Boussalis2019; Schoonvelde et al. Reference Schoonvelde, Brosius, Schumacher and Bakker2019). All of this is not to say, of course, that such an approach is always foolproof (Benoit, Munger, and Spirling Reference Benoit, Munger and Spirling2019) or that there may not be other ways of investigating the simplicity of populist communication. Discourse analysis is certainly one, while looking at the integrative complexity of political language (as Tetlock Reference Tetlock1983 and others have done) is another. However, our approach does seem in line with the broad understanding of simple language in the theoretical work on populism (recall, for example, the references to populists not using “bureaucratic jargon”) and the empirical studies conducted to date (not to mention the media discussion of this issue in relation to Trump).
To conclude, the existing scholarship on populism leads us to expect that right-wing populists will use simpler language than their competitors, a claim that is partially supported by work to date. However, ours is a tentative expectation for a couple of reasons. First, the few relevant empirical studies do not tell us whether simpler language is a consistently shared and distinguishing feature of populist leaders across countries, as the theoretical literature on populism has long asserted. Second, since we know the language of mainstream political leaders has become simpler over time (Lim Reference Lim2008; Cortelazzo and Tuzzi Reference Cortelazzo and Tuzzi2008), it could be that the gap between elite and populist language has also reduced, thus making assertions about comparative populist simplicity outdated. Since such assertions are overwhelmingly predicated on cases of right-wing populists (which have been, by far, the dominant ideological types of populists in western democracies since 2000) rather than left-wing ones, we focus in our study on the linguistic simplicity of recent right-wing populists compared to their principal mainstream rivals in four major democracies. In the next section, we present these cases and explain the data and methods we have used to investigate them.
Cases, Data and Methods
Our four right-wing populist cases are Donald Trump (United States), Nigel Farage (United Kingdom), Marine Le Pen (France), and Matteo Salvini (Italy). All have been recent party leaders or presidential nominees in their respective countries.Footnote 3 They have also all been widely treated in the literature as right-wing populists (e.g., Müller Reference Müller2016; Finchelstein Reference Finchelstein2017; Mudde Reference Mudde2019; Rooduijn et al. Reference Rooduijn, Kessel, Froio, Pirro, Lange, Halikiopoulou, Lewis, Mudde and Taggart2019). In the US and France, we compare Trump and Le Pen with their main opponents, Clinton and Macron, during those countries’ respective 2016 and 2017 presidential election campaigns.Footnote 4 In the two parliamentary systems, Italy and the UK, we are not faced with a two-horse election campaign between non-incumbents as in the United States and France in 2016 and 2017. For this reason, we compare those countries’ main right-wing populist party leaders of recent years, Farage of UKIP and Salvini of the Northern League, with the most prominent leaders from both centre-left and centre-right over similar time periods: in the UK, David Cameron (leader of the Conservative Party and, for some of the analysis period, prime minister) and Ed Miliband (leader of the Labour Party);Footnote 5 in Italy, Matteo Renzi (leader of the centre-left Democratic Party and, for some of the analysis period, prime minister) and Angelino Alfano (leader of the New Centre Right).Footnote 6 We are thus able to test the linguistic simplicity claim across four countries, three languages (English, French, and Italian), and two political systems (presidential and parliamentary) with different types of right-wing populist and other leaders. If right-wing populists really do consistently use simpler language than their rivals, we should find this to be the case irrespective of those differences.
Corpus Compilation
Table 1 sets out the main characteristics of our corpus in terms of speakers, the time period from which we took speeches for each politician, the total subcorpus size, and the mean length of speeches (further details about the speeches and the methods used to analyze them are available in online appendix A).Footnote 7 The number of tokens in table 1 refers to the total number of running words, as opposed to word-types, a term used in linguistics to refer to the list of all the different forms included in a corpus. For example, in any text we will find many repetitions (i.e., word-tokens) of single word-types such as “the,” “a,” “of,” etc.
Corpus composition
Our dataset, totaling 1,032,216 tokens, ensures good comparability and representativeness for several reasons. First, our speakers within each country case are comparable. In France and the United States, we compare non-incumbent presidential candidates. In the UK and Italy, we compare party leaders. Second, our time spans are comparable. As table 1 details, the speeches in each country occur in the same general period (i.e., we are not comparing leaders from different eras). Third, our text types within each country case are comparable: in the United States and France, they are monologues delivered during the election campaigns by both candidates. In the UK and Italy, they are also monologues and are delivered to what we can usually consider friendly audiences at events such as party conferences, rallies, campaign meetings, etc. We have deliberately not selected debates, interviews, and press conferences because the dialogical nature of these means the speakers can be considered more like “dancing pairs.” For example, speakers may be led by their interviewer/opponent to speak about topics in certain terms and the interruptions and turn-taking strategies of these exchanges affect linguistic complexity. Similarly, we deliberately avoided using parliamentary speeches since speakers are constrained by etiquette and allotted time (Cortelazzo Reference Cortelazzo, Holtus and Radtke1985; Van Dijk Reference Van Dijk and Bayley2004; for an overview of parliamentary discourse genres, see Ilie Reference Ilie, Tracy, Ilie and Sandel2015). Our subcorpora are also comparable in terms of size as approximately 100,000 tokens were transcribed for each leader.
Finally, it is worth acknowledging that speeches are probably written in advance by a team of ghost writers (with varying degrees of input from the leader). However, unlike manifestos for example, there is only one individual who actually utters the text in front of the public and takes on the responsibility for what is said. In other words, the public does not have access to “Trump’s original and true discourse” in the campaign: they are only familiar with his public linguistic image, which they get from his publicly uttered words. So, if we want to understand whether populist leaders use simpler language in public than other leaders, it is not relevant whether the words are all their own (see also Wang and Liu Reference Wang and Liu2018, 306-307).
Methods
Our aim is to assess the linguistic simplicity of the ten cases, with linguistic simplicity understood as the extent to which a speaker’s language is easy to comprehend. We have used a broad set of measures to analyze this. In addition to readability scales, we tested each subcorpus for lexical richness, lexical density, and the percentage of words that are considered difficult in particular languages. To be clear: we are not proposing any new methods here, but are using an array of standard, well-established methods that have long been used by linguistics researchers (and, occasionally, political scientists) to assess the simplicity of language. What is innovative is, first, the comparable and homogenous corpora we have constructed and, second, the range of measures we use, which goes well beyond what has been deployed before to assess the comparative linguistic simplicity of populists. Table 2 lists these measures along with the basic rationale behind each. We discuss them in greater detail in the following paragraphs.
Linguistic simplicity measures
We calculated the readability of texts using a range of indices. For our U.S. and UK cases, we used the Flesch Kincaid Reading Ease (FK) and Flesch Kincaid Grade Level (FKGL) indexes (Kincaid et al. Reference Kincaid, Robert P., Richard Lawrence Rogers and Brad1975) and then cross-checked our results with a series of other common readability measures.Footnote 8 FK assigns scores from 0–100 to texts, with higher scores indicating greater readability. It divides this scale into categories spanning from “very difficult” (0–29) to “very easy” (90–100). FK and FKGL were devised for English and are not entirely suitable for French and Italian. For the former, we used the Kandel-Moles (KM) index, designed specifically for French (Kandel-Moles Reference Kandel and Moles1958), and supplemented it with LIX, which has been proven to perform well on Western European languages.Footnote 9 Finally, for our Italian subcorpora, we used a readability index called Gulpease, which was developed for Italian and is recognized as the standard for that language (Lucisano and Piemontese Reference Lucisano and Piemontese1988). We checked the readability measures of our Italian subcorpora by also calculating the LIX score.
As Schoonvelde et al. (Reference Schoonvelde, Brosius, Schumacher and Bakker2019, 5-6) discuss, FK and similar tools have long been used to analyze political speeches. While common, their suitability for assessing the simplicity of contemporary political language has been questioned recently by Benoit, Munger, and Spirling (Reference Benoit, Munger and Spirling2019), on the grounds that such measures cannot distinguish between texts that are “clear” (which is of course positive) as opposed to “dumbed down” (which is negative). As they argue (492): “the fact that we might communicate the same complex content, but in shorter words and sentences that require less processing by the reader, is almost certainly a good thing,” but could be construed by measures like FK as “appealing to a less educated audience.” To address this problem, they propose an interesting new method using crowdsourcing to conduct comparisons of thousands of text snippets as part of an analysis of what they term the “political sophistication” (i.e., complexity) of U.S. State of the Union speeches. Nonetheless, as they acknowledge, this only slightly improves on Flesch Reading Ease (Benoit, Munger, and Spirling Reference Benoit, Munger and Spirling2019, 501). Moreover, given the greater number of languages (three) and country cases (four) we have to contend with in our study, their approach would almost certainly be unfeasible for us.
We complemented our readability analyses with a number of other widely used linguistic simplicity measures. The first is lexical richness, which is based on the premise that the higher the repetition rate of word-types in a corpus, the easier the language being used, since lexical variation, in addition to increasing difficulty per se, may also imply a broader range of contents. As an additional check of lexical richness, expressed as the Type/Token Ratio (TTR) and percentage of hapax legomena (i.e., words occurring only once in the corpus), we calculated the lemma/token ratio (LTR). This is because, when we refer in everyday speech to the “rich vocabulary of a speaker,” we tend to mean the unusually large number of different lemmas used (i.e., the basic forms for a paradigm, such as “love” as the basic form for “loves,” “loved,” “loving,” etc.) and not the great variety of inflected forms (Granger and Wynne Reference Granger, Wynne and Kirk1999). The second is lexical density, expressed as the share of grammar words in a text, i.e., words that are used primarily to build syntactic structures (conjunctions, determiners, prepositions, and pronouns) and content words, i.e., words conveying meaning (adjectives, adverbs, nouns, and verbs). If a text conveys more information than another of equal length, then it is more difficult to understand.
Finally, we have assessed how difficult the lexis of our corpora is. While the main simplicity measures are based on word length and word variation, it can of course be the case that shorter words used by a speaker are more difficult to understand for the average citizen (e.g. “ersatz” is shorter, but considered harder to comprehend, than “substitute”). To investigate this, we have used indices based on the presence of words that have been evaluated as accessible to the general public. In the case of English, we used the Dale-Chall score (DC) while, for Italian, we used Corrige! to calculate the percentage of words considered as belonging to the Italian basic vocabulary (Vocabolario di Base; henceforth: VdB).Footnote 10 No appropriate vocabulary lists are available in French, so we omit the “difficult words” analysis for that case.
All of the measurements detailed here were replicated for each leader’s subcorpus to assess whether the language of right-wing populist leaders is simpler than that of their principal mainstream rivals. To be clear: while we use the same types of measures on all speakers, it only makes sense to compare between speakers within the same language and national political culture, rather than to do so across languages and cultures. In other words, we are able to say if Le Pen uses simpler language than Macron, but not if she uses simpler language than Clinton. Indeed, even comparing between English-language speakers like Trump and Farage is problematic because the U.S. and UK have different traditions and habits when it comes to political discourse. In any case, this is not an issue for us given that our objective is to understand whether right-wing populists use simpler language than their major national competitors during the same period. Of course, if we find that they do so in different languages and different political cultures, then that provides support for the universal claims regarding the simple language of populists.
A possible objection to our approach is that putting together some twenty or more speeches delivered by the same political leader may give us an idea of how that leader speaks in general but, in the real world, voters listen to individual speeches and no-one listens consecutively to 100,000 words. This is certainly true, however the usual solution to offset the impact of text size, i.e. analyzing a 500 token “moving window” (Wang and Liu Reference Wang and Liu2018; Ondelli and Nadalutti Reference Stefano, Nadalutti and Palumbo2017) may be subjected to the same criticism: voters do not listen to 500 token chunks, and certainly not to chunks including truncated sentences. Nonetheless, in order to assess the statistical reliability of our subcorpora and confirm that our overall readability values are not misleading due to text length differences, for each speech we calculated readability scores that are unaffected by text size and conducted standard tests to assess the statistical significance of our findings (refer to online appendix B). We also used the Shapiro-Wilks normality test to check that the distribution of readability scores within each subcorpus was normal and carried out T-tests to determine whether the means of our sets of data were significantly different from each other. These robustness checks confirm the results presented in the next section (refer to online appendix B). Finally, it is true that we cannot consider the effect of factors such as speech speed and intonation, as well as pauses, on simplicity. However, this applies also to other studies of the simplicity of political language. What we have done, as explained, is to carefully assemble a sufficiently large set of comparable speeches for each of our four country-cases, which we examine with a wider range of well-established linguistic measures than has been used before, to analyze the comparative simplicity of populist and non-populist language.
The Comparative Linguistic Simplicity of Right-Wing Populists
Table 3 presents the overall results of our analysis, conducted using the methods discussed above (further details are available in online appendix A). Although there are differences between them, the four right-wing populists taken as a group do not meet our expectation that they always use simpler language than their mainstream rivals. In the remainder of this section, we discuss our principal findings country-by-country before considering the overall results, and their implications, in the concluding section.
On which measures are right-wing populist leaders linguistically simplest?
Note: Ticks indicate that right-wing populists are linguistically simpler than their main rivals (Clinton in United States, Macron in France, Cameron and Miliband in UK, Alfano and Renzi in Italy). Grey boxes indicate that right-wing populists were not the simplest politician on a particular measure amongst those we analyzed in their country. As indicated by the “N/A” box, there is no measure for assessing difficult words in French.
United States
As table 4 shows, we find that Trump’s language is generally simpler than Clinton’s. However, the gap is much narrower than in studies of their respective speeches during the primaries (Oliver and Rahn Reference Oliver and Rahn2016; Kayam Reference Kayam2018) and he is not simpler on all of our measures.
The linguistic simplicity of Trump and Clinton
Note: For “readability,” higher FK scores and lower FKGL scores denote greater simplicity; for “richness,” lower TTR, LTR, and Hapax scores indicate greater simplicity; for “density,” lower content (and higher grammar) scores indicate greater simplicity; for “difficulty,” lower DC scores indicate greater simplicity.
First of all, regarding readability, the two candidates are very close to one another with Trump at 72.4 and Clinton at 71. Both scores thus fall within the “fairly easy” band on the FK scale. On the FKGL scale, Trump’s speeches are rated as comprehensible to a sixth-grader, while Clinton’s require a seventh-grade level of education. Since we know that the readability score is based on sentence and word length, it is worth looking more closely at those two measures.Footnote 11 Specifically, we find that Trump uses slightly more words that are longer than three syllables (10.97% versus 10.75%, and an average word length in syllables of 1.44 versus 1.43), but Clinton uses longer sentences (on average 15.02 words per sentence compared to 12.55 for Trump).
Second, we find that Clinton’s speeches are lexically richer (albeit only slightly) than those of Trump, since TTR, LTR, and hapax values are greater in her corpus. In other words, when Clinton speaks, she uses both a wider range of morphological variants of the same words (e.g., love, loves, loved, loving, lover, etc.) and a greater number of distinct words (e.g., love, hate, passion, fear, etc.). This might indicate that, given the same total amount of words, either she speaks about a greater variety of contents (requiring different words to be expressed) or, when dealing with the same contents, she uses a broader range of synonyms, in both cases increasing the complexity of her language.
The only measure according to which Trump’s language may be considered more complex than Clinton’s is lexical density, since he uses more content words than her. Finally, if we look at the difficulty of words used, we see that Clinton has a higher Dale-Chall score, indicating greater complexity. Nonetheless, both speakers fall within the same range on that scale since values below 4.9 indicate that the text can be easily understood by an average 4th-grade student.
Overall, the U.S. results are consistent with the theoretical claims about right-wing populists using simpler language. However, as noted, the differences between the two candidates are relatively small.
France
If the U.S. results are slightly surprising, those from the French case are much more so. Table 5 details how, on all measures, Marine Le Pen’s language during the 2017 presidential campaign was markedly more complex than Emmanuel Macron’s.
The linguistic simplicity of Le Pen and Macron
Note: For “readability,” higher KM scores and lower LIX scores denote greater simplicity; for “richness,” lower TTR, LTR, and Hapax scores indicate greater simplicity; for “density,” lower content (and higher grammar) scores indicate greater simplicity.
To test readability, we used both the KM index and the LIX scale: according to both, Le Pen is more difficult than Macron. On the KM 0-100 index, Le Pen is clearly distant from Macron with a score of 57 compared to his 65 (hence, they are in different KM bands, with Le Pen rated as “difficult” and Macron as “standard”). According to LIX, Le Pen is again the more complex speaker, although the distance between the two is less. Regarding the factors determining the readability scores, Le Pen uses slightly longer sentences than Macron (on average her sentences are 22.27 words long compared to Macron’s 21.51). We also find that Le Pen is less simple than Macron as regards lexical richness (with approximately a 3-percentage point gap in both TTR and hapax values and almost 2 percentage points in LTR), lexical density and average word length (4.76 letters vs. Macron’s 4.48).
The French results are thus very clear: contrary to what the literature on populism would lead us to expect, Le Pen’s language is not at all simpler than that of her opponent, Macron. Moreover, unlike the U.S. results (where the gap between Trump and Clinton was small), the differences between the two French politicians are much greater on all our measures. In the conclusions, we discuss some of the reasons why this may be the case.
United Kingdom
While not as clear cut as the French results, those from the UK also show that right-wing populists do not always use simpler language than their competitors.
As table 6 details, in terms of readability, Nigel Farage is clearly more difficult than both Ed Miliband and David Cameron. On the FK scale, his speeches place him at 63.2, while Cameron and Miliband are at a considerable distance on 71.8 and 72.9, respectively. These results put Farage in the lower reaches of the “standard” FK band, with both Cameron and Miliband instead rated “fairly easy.” In terms of the FKGL, Farage is considered comprehensible to a (United States) tenth grader, Cameron to a seventh grader, and Miliband to a sixth grader. The main reason for Farage’s lower score is the length of his sentences compared to the other two politicians. While Miliband’s sentences are on average 13.99 words long, and Cameron’s are 15.49, Farage’s are a remarkable 24.61. However, while his sentences are longer, the words they are composed of tend to be slightly shorter than those used by Miliband and Cameron (the share of words over three syllables is 9.54% for Farage, 9.66% for Miliband and 9.72% for Cameron, with 1.40, 1.42, and 1.41 average syllables per word respectively).
The linguistic simplicity of Farage, Miliband, and Cameron
Note: For “readability,” higher FK scores and lower FKGL scores denote greater simplicity; for “richness,” lower TTR, LTR, and hapax scores indicate greater simplicity; for “density,” lower content (and higher grammar) scores indicate greater simplicity; for “difficulty,” lower DC scores indicate greater simplicity.
Moving on to the other measures, Farage is reasonably similar to the other two politicians and is certainly not noticeably simpler. For lexical richness, he scores lower than Cameron, but higher than Miliband in all three indices. Regarding lexical density, Farage has the lowest score for content words, but the difference compared to Miliband is less than one percentage point. Finally, when we look at difficult words used, we find that, while all three are in the same Dale-Chall band (between 4 and 5), Farage is the most complex of the three. Overall, therefore, our results show that it is the center-left leader Miliband, not the right-wing populist Farage, who uses the simplest language according to three of our four measures.
Italy
Our findings for the Italian case, set out in table 7, present a more mixed picture than those for France and the UK.
The linguistic simplicity of Salvini, Renzi, and Alfano
Note: For “readability,” higher Gulpease scores and lower LIX scores denote greater simplicity; for “richness,” lower TTR, LTR, and hapax scores indicate greater simplicity; for “density,” lower content (and higher grammar) scores indicate greater simplicity; for “difficulty,” lower non-VdB scores indicate greater simplicity.
According to both readability indices, Matteo Salvini’s language is simpler than that of Matteo Renzi and Angelino Alfano. His speeches are ranked sixty-two according to Gulpease, within the band that Italians with a middle-school education ought to find easy. By contrast, Alfano’s are close to being considered almost incomprehensible by people with that education level. The main reason for Salvini’s higher readability score is the shorter length of his sentences: 17.22 words, compared to Renzi’s 22.52 and Alfano’s 33.78. The same pattern is discernible for average word lengths in characters of the three leaders, with Salvini at 4.51, Renzi 4.70, and Alfano 4.80. We find a similar distribution (and for the same reasons) when we calculate readability using the LIX test.
While readability indices give a fairly clear result of greater right-wing populist simplicity in Italy, all our other measures present a different account. Salvini’s speeches are both lexically richer (albeit slightly, as shown by the TTR, LTR, and hapax values) and more difficult (with 8.7% of his words not belonging to the Italian basic vocabulary, compared to 7.6% for Renzi and 6.8% for Alfano). We also see that Salvini is lexically denser than Alfano, and only slightly less so than Renzi. Overall, therefore, we can say Salvini’s shorter sentences make him easier to understand at syntactic level, but that he displays greater lexical inventiveness, thus adding to the complexity of his language.
Conclusion
In this study, we have analyzed the linguistic simplicity of four right-wing populist leaders compared to their principal mainstream competitors in the United States, France, the United Kingdom, and Italy. Taken together, our cases do not show a clear pattern in support of the claim that right-wing populists use simpler language than their rivals. Trump is the populist leader most in line with the theoretical literature. Nonetheless, while he is generally simpler than Clinton, ours is not a “Trump speaks at the level of a fourth grader” story. Rather, we find that Trump and Clinton fall within the same bands on the FK and Dale-Chall scales and are only separated by one grade on the FKGL. In Italy, Salvini is simpler than Renzi and Alfano according to the readability indices but is not the simplest on any of the other three measures. Finally, the results from France and the UK are clear-cut, but not in the way we anticipated. Farage is generally less simple than Miliband and Cameron, while Le Pen is constantly less simple than Macron on all our measures. The long-standing belief that right-wing populists are characterized by their use of simpler language is thus not borne out by our study.Footnote 12
How can we explain this counter-intuitive finding? First, we are open to the theoretical possibility that if we had examined different outlets other than speeches, we might have found our right-wing populist cases used simpler language than their rivals (although this remains an open empirical question). However, in the light of the methodological issues we raised earlier regarding debates and interviews, and the fact that speeches are the most traditional form of public political language, we are confident that studying them is sufficient to test the claim that populists are characterized by their use of simpler language than their opponents.Footnote 13 Second, we should also acknowledge that our findings may depend on the period analyzed. We chose four recent cases of right-wing populists and so it is possible that, had we looked at an earlier decade, we might have found results more in line with the literature on populism. There are plausible reasons to think this. As mentioned earlier, we know that mainstream politicians in some countries have simplified their language compared to previous eras. Lim (Reference Lim2008), for example, shows how the readability level of U.S. presidents’ speeches has increased significantly over time. Similarly, Cortelazzo and Tuzzi (Reference Cortelazzo and Tuzzi2008) found that there has been an overall simplification of Italian Presidents’ annual New Year’s speeches since the 1980s.Footnote 14 However, we do not know whether the rate (or direction) of simplification has been different for right-wing populists compared to other leaders. It could be the case, for example, that, as right-wing populists have moved from being pariahs to potential members of governments, they have not further simplified their language, or they have done so at a slower rate than mainstream politicians, or they have even made it more complex in an effort to sound suited to ministerial roles. A study of the linguistic simplicity over time compared to their various rivals of right-wing populist leaders such as Timo Soini of the Finns Party and Heinz-Christian Strache of the Austrian Freedom Party (who both went from the margins to government) would be useful in this regard.
Another possible explanation is that, as mainstream political elites have followed the advice of professional communications advisers and reduced the linguistic complexity of their speeches, right-wing populists in some contexts have chosen to act differently, in order to appear less coached and more authentic. Farage’s long sentences make him stand out from Miliband and Cameron and (presumably) add to his “man holding court in the pub” image. Similarly, as a French nationalist who opposes globalization and its alleged cultural homogenizing effects (Betz Reference Betz, Kriesi and Pappas2016, 76), Le Pen and her advisers may see an advantage in not imitating the Anglophone political language trends that, by contrast, Macron has embraced. After all, both Farage and Le Pen espouse a traditional image of what being British or French means (as opposed to “European” or “cosmopolitan”). Both have also praised mainstream leaders of the past like Margaret Thatcher and General De Gaulle, suggesting that their successors are inferior and unworthy of those national heroes’ legacies. Opting for the linguistic setup of their national rhetorical traditions as opposed to slogan-based simpler communication techniques derived from the United States model may thus be considered useful for these right-wing populist leaders in the UK and France.Footnote 15 As we have shown, they do distinguish themselves linguistically from Cameron, Miliband, and Macron. The surprise is that they do so not by being simpler than these elites, but more complex.
What do these findings mean for our understanding of populism? The idea that populists portray themselves as standing with ordinary people against the elites is at the heart of the concept and scholars have long claimed that populists use simpler language than those elites in order to drive home their simplistic messages and reinforce their positions as men and women of the people (Canovan Reference Canovan1999; Zaslove Reference Zaslove2008). Our research suggests, however, that we need to distinguish simple language from populist arguments that are perceived by scholars as simplistic. One does not automatically accompany the other. Similarly, the automatic pejorative association of populists (or any other politicians) with simple language should also be reconsidered. As Benoit, Munger, and Spirling (Reference Benoit, Munger and Spirling2019) rightly argue, there is a difference between simple language that is clear and simple language that is dumbed down. For example, The Catcher in the Rye and The Old Man and the Sea score 3.4 and 3.9 respectively on the Flesch Kincaid Grade Level, while Harry Potter and the Philosopher’s Stone and The Da Vinci Code score 4.7 and 5.8.Footnote 16 But few would argue that the greater linguistic simplicity of the first two novels indicates they are more dumbed down than the latter.
From our study, we also now know that voters are not necessarily choosing between mainstream politicians who speak to them in higher registers and populists who do so in simpler language. So, what is it—if anything—that distinguishes the language of populists from those of their competitors and what is it in their language that establishes them as the defenders of the people against the elites? One possibility is that the simplicity of the concepts or the “integrative complexity” (Tetlock 193, Reference Tetlock1984) differentiates populist speeches from those by mainstream politicians. We see this as a very fruitful avenue for future research (and is complementary, rather than in opposition, to what we have done). Our study also suggests that, if we wish to understand the defining linguistic features of populists, we need to look beyond simplicity and consider the choice of vocabulary and rhetorical instruments used to convey recurrent contents. For example, in what ways and with what frequency do populists in different languages refer to their conceptual pillars of “the people,” “the elites,” and (in the case of right-wing populists) “the others” compared to the major mainstream politicians in their countries?
Finally, our findings should make us question other aspects of the received wisdom about populists. The claim of simpler language has long gone hand-in-hand with the idea that populists are vulgar figures who appeal to people’s baser instincts, by “lowering the level of political discourse” (Moffitt Reference Moffitt2016, 60-61). Again, however, the notion of populists being more vulgar may seem intuitively convincing but is one that—just like linguistic simplicity—needs to be operationalized and tested. It may be clearly true in some cases, but not others.Footnote 17 Similarly, if right-wing populists do not always use simpler language, then the related negative assumption that simple and vulgar language must be appealing to those who vote for them also needs to be reassessed. The underlying message of the Boston Globe story and much of the research literature is that right-wing populists deliberately speak like fourth graders and their (simple) supporters lap it up. Our research shows that while this may be a convenient and even comforting idea, the reality—like populist language itself—is likely to be rather more complex.
Supplemental Materials
Appendix A. Subcorpora Composition and Simplicity Analyses
Appendix B. Statistical Checks
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592720002418.
