The data hereby presented refers to the Twitter activity of the candidates to the European Parliament and of the most influential party figures/accounts from the main coalitions of France, Germany, Ireland, Italy, Spain and the United Kingdom.
A look at how the European Twitter-sphere reacted to Marine Le Pen’s anti-globalization message.
A great deal of water has passed under the bridge of European politics since the two-day Lyon rally of February 4-5, in which Marine Le Pen launched her campaign for President of France. Now that independent centrist Emmanuel Macron has settled into the Elysée Palace, the memory of a presidential campaign in which France’s role in the globalized world was put in question might seem far.
In her 144 commitments Le Pen pledged, among others, to withdraw from the euro, hold a referendum on EU membership, implement “intelligent protectionism”, and deter immigration. With the promise to “give France back its freedom”, Le Pen’s manifesto took a strong stance against the European integration process, offering French people a shield against globalization forces. In her words: “The divide is no longer between the left and right but between patriots and globalists.”
For forty days–starting on 4 February until 15 March–we followed Marine Le Pen’s campaign on Twitter, investigating if and how such a divide shaped the online discussion of French elections, in France as well as in other EU member states. More precisely, we collected all tweets sent by or to the leader of the Front National, plus all tweets that simply mentioned her. All in all we collected 4,185,917 such tweets, which we first processed so as to estimate the location of the sender. We then discarded all tweets sent from outside the EU as well as those with an undetermined location. Eventually we were left with nearly 1.5 million “European” tweets. As second step, we took a closer look at conversations originating from France.
Political content on Twitter comes in many forms: campaign content, messages booing or cheering the candidates, commentaries, “flames”, news sharing, but also puns, irony, memes and satire. Conversations about the issues–meaning here tweets mentioning precise policy options–are more often the exception than the rule.
Figure 1: Wordcloud of the 100 most frequently used hashtags in France
French presidential elections were no exception: for a quick look at 40 days worth of tweets about Marine Le Pen, in Figure 1 we have plotted a word cloud with the 100 most frequently used hashtags in France. Judging from this figure, we might not notice any particular sign of Euroscepticism: the #UE (Union Européenne) hashtag hardly makes to the top 100, and the same is true for #frexit, #schengen and #CETA.
[Among French users discussing Le Pen’s campaign], roughly one tweet out of ten expresses an opinion on a globalization-related topic.
To understand the extent to, and the ways in which, French Twitter users discussed Le Pen’s stance towards globalization in general, and the European integration process in particular, we needed to bring to the surface tweets explicitly expressing attitudes or opinions. By means of natural language processing (NLP), we built an algorithm to separate “opinion” tweets from news recast or merely descriptive tweets. In this way we eliminated 300,159 tweets of the latter kind (40.1% of the French sub-sample). We then filtered the remaining tweets through a second algorithm built to recognize all tweets expressing opinions towards “globalization”: here we used a broad criterion that included in this category all views expressed with regard to a wide range of topics such as immigration, trade, sovereignty, European integration, the European common currency, and so on. As Figure 2 shows, roughly one tweet out of ten expresses an opinion on a globalization-related topic.
Figure 2: Globalization-related opinion tweets vs other opinion tweets – France
This result is particularly interesting since during the electoral campaigns each single issue (although broadly defined) has to compete with a great variety of other topics: the high salience (11% of all attitudinal tweets) of the topic could be seen as showing Le Pen’s relative success in setting the agenda of the electoral debate.
Before further delving into those globalization-related tweets, we took a look at what happened outside France, and tried to establish if a similar pattern emerged in our set of “European” tweets.
France’s presidential election was followed closely by international commentators and observers as a key political event in which populist forces could gain political momentum. Le Pen’s possible victory, although unlikely, would have seriously challenged the resilience of the European integration process. As a consequence, online discussions related to Le Pen, although dominated by French users as expected, featured a non-negligible share of tweets sent from other European countries (Figure 3).
Figure 3: Country distribution of tweets (EU member states only)
A quick look at the hashtag cloud of European tweets (see Figure 4 below) provides support to this view: the frequency of the #eu hashtag, and of the #brexit associated to #frexit, as well as the emphasis placed on foreign political figures such as #wilders, #trump or #putin, are hints of the presence of some extent of horizontal Europeanization (and even internationalization) around this topic–as we also observed for the Italian constitutional referendum.
Figure 4: Wordcloud of the 100 most frequently used hashtags in EU member states, excluding France.
As we did for the French conversations, we developed a set of algorithms to disentangle discussions and comments about globalization and European integration from the generality of tweets originated in Europe. We used NLP techniques on the subset of tweets in English–the most used language in our sample outside French–in the hope of capturing the attitudes of a koine of twitter users that uses English as lingua franca when tweeting about European matters. As a first step, we broke down our European subsample of 198,568 tweets into attitudinal tweets on the one hand (58,84%) and news recast, descriptive or link-sharing tweets on the other (41,16%). We then processed the 116,831 attitudinal tweets to separate those expressing opinions on globalization-related matters from the rest.
Le Pen had a hard time trying to persuade users to embrace her views on the European Union, free trade, the Schengen area or the common currency.
As Figure 6 shows, two out of three attitudinal tweets about Marine Le Pen’s campaign are somewhat related to “globalization” in a broad sense, ranging from discussions about the common currency to takes on immigration, and so on. It is therefore plausible that the prospects of success on the part of a anti-globalist, anti-European candidate triggered processes of both horizontal and vertical Europeanization, namely a significant degree of attention paid to political events in other European countries (horizontal) as well as the presence of a debate on the EU itself (vertical).
Figure 6: Globalization-related opinion tweets vs other opinion tweets – EU member states
The most immediately visible side of an electoral campaign on Twitter is the “official” one, i.e. the one in which party slogans are conveyed, candidates share campaign press releases or endorsements, events like rallies or television shows are documented. We tried to bring to the surface a less visible layer of the debate composed of comments, attitudes, praises and criticism about candidates and electoral issues. At this level, we were able to disentangle, both for French and European tweets, conversations and discussion that shape what can be defined a divide between “patriots and globalists”.
Outside France, almost three out of four tweets express a negative sentiment towards globalization, the EU or related issues.
As a final step, we tried to determine on which side of this divide Twitter users stood. We did this by NLP aimed at assessing tweets’ attitude towards the EU and globalization, in conversations related to Le Pen’s campaign. As Figure 8 and 9 show, our results differ significantly when considering French tweets and tweets from other EU countries. In France the share of tweets questioning the Front National’s positions on globalization-related issues seem to prevail (56,7%, see Figure 7). As far as online conversations are concerned, Le Pen hence had a hard time trying to persuade users to embrace her views on the European Union, free trade, the Schengen area or the common currency. (It is perhaps no coincidence that she later dropped the idea of Frexit and a return to the French franc from her platform altogether).
Figure 7: Sentiment towards EU and globalization – France
Outside France, on the other hand, the picture is reversed. Users who tweeted about globalization or the EU seem to embrace Marine Le Pen’s stance with much more enthusiasm: almost three out of four tweets express a negative sentiment towards globalization, the EU or related issues (see Figure 8).
Figure 8: Sentiment towards EU and globalization – EU member states
In explaining this result, the high share of tweets coming from the United Kingdom should be taken into account, as it suggests some kind of “Brexit effect” that should be investigated more thoroughly. Nevertheless the contrast with the French situation remains striking, and seems to hint at a dimension of horizontal Europeanization mainly driven by anti-globalist forces, in which the online voices of Europhiles resound far less than those of the Eurosceptics.
Photo Credits CC David Oranje