The Guest Blog

Guest blog post by EU-Reputation.EU

  1. Timeline
Timeline - SOTEU -
Timeline – SOTEU –

The graph shows the timeline of the speech and where the volume of tweets hit their cruising speed. The peaks correspond to the different announcements made by Mr Juncker and the reaction to them. Because the tweet rate decreased shortly after each announcement was made, it shows that the discussions are mostly held by journalists.


II. Semantic analysis

  1. Keywords analysis

Disclaimer : Only tweets in English and French were analysed in this section


Regarding the overall number of mentions through articles and tweets, we notice that there is a significant representation of the Brexit issue. Other than Brexit, the keywords are mainly neutral. Following the theme of the Brexit issue, let’s focus on what the keywords in the UK were:


We can see that the Essex story and Brexit are more prevalent there.


Concerning the overall sentiment of Mr Juncker’s speech, we analysed the difference between the positive and negative keywords.feeling


Once again, we see that the crisis that affects the Union and Brexit predominantly relate to negative discussions. Focusing deeper in the UK, we can see that there is a stronger negative feeling there.feeling-camembert1


If we exclude neutral mentions and tweets (news headlines, for instance), we see a much bigger difference :


NB : Tonality algorithm are not 100 % trustworthy and could be twisted. However comparing 2 different results from the same algorithm is a good indicator. :


  1. Top Tweets

At the end of the speech, here were the Top 5 Tweets, ranked by number of RTs :


We clearly see there that the Top 5 tweets are tweets sent by EU institutions. Interestingly, there is no journalist, politician or citizen featured. This shows clearly that, with regards to discussions, the power of the EU institutions is even bigger than the power of the people who run them.


By taking into account the after-speech and the debate that was held in the European Parliament (until 2:30 PM), the ranking evolves the following way :


Marine Le Pen is the only politician in this top 5 having local political responsibilities, showing that she has a lot more engagement than others. She even has more engagement on this subject than any of the EU institutions or Jean-Claude Juncker himself. Guy Verhofstadt, a well known politician in European affairs, is also featured there.


III. Demographics

  1. Gender analysis

Nb : Gender repartition is based on first names


There were more men commenting and reacting to the State of the Union than women.


  1. Interest analysis

Nb : Centre of interests are defined by scanning account description and tweet activity


Unsurprisingly, journalists, people engaged and interested in politics as well as business are well represented. Further breakdown by profession also shows that a Brussels bubble specialised in consulting, journalism or entrepreneurship exists.


Quite logically, privacy issues and 5G were more discussed in the tech community whereas the Brexit was very central in the economic community.

From the political community, EU’s existential crisis was a key topic.


  1. Geographic analysis


At a national level, we can see there were significantly more participants taking part from the United Kingdom, followed by Belgium and France.

Nb : Countries are obtained through GNIP datas for Twitter and journalist/blogger website (Source : Brandwatch)

Nb : les pays sont obtenus via les données fournies par GNIP pour Twitter et par le site du journaliste / blogueur (Source : Brandwatch)
Nb : les pays sont obtenus via les données fournies par GNIP pour Twitter et par le site du journaliste / blogueur (Source : Brandwatch)

This is quite interesting because if we compare it with the global activity about the speech (the total volume of tweets), we notice that Belgium, and therefore Brussels was the most active country by far :


This shows that the EU stakeholders are more active than others. But we also notice that the tweet activity were mostly concentrated in EU Capitals, with 2 exceptions : Catalonia and Scotland, two regions where there are strong discussions about the EU.


Concerning the language issue, we can see that English remains the most spoken language in EU discussions, followed by French and Spanish.


Nb : Language repartition via Visibrain


  1. Stakeholder analysis


The President of EU Commission, Jean-Claude Juncker, is by far the most mentioned or retweeted user by other stakeholders. Nevertheless, all users in this list are from the EU bubble except Marine Le Pen. There is no big media or civil society represented there. If we cross reference this with Social Media analysis methods, we get this map (before the speech). We can clearly see that, before the speech discussions are polarised around EU institutions and divided by language communities.




The same analysis, after the speech (at 1 PM) shows that the European bubble (in blue), was commented in several different languages. Around the bubble we can see local stakeholders speaking their own languages and from different kind of opposition standpoints (Spanish far-left, for instance).



Jean-Claude Juncker’s speech has been mainly shared and commented within the European bubble, with, this year, a British influence mainly because of Brexit. As Mr Juncker recalled it, one of Europe’s difficulty is to deliver a positive image at national and local levels. We can see through Marine Le Pen’s tweets that national stakeholders have a lot more visibility than the European ones. This is also verified through the presence of other local stakeholders such as accounts related to the Spanish far-left.


And thus this is the main difficulty for the EU; to be able to speak to people who are not listening to you.


Méthodology :

This study has been done with Brandwatch (gathering all social medias and news websites) and Visibrain, media monitoring tool (only Twitter).


  • On Brandwatch, keywords were “SOTEU OR Juncker OR (Discours Etat union) OR (State UNION speech)” only on the 14/09.
  • On Visibrain, the only keyword was “SOTEU” on the last 24 hours.

Every tool is mentioned on every chart legend.


Contact about the study :


Nicolas Vanderbiest

PhD Assistant at UCL

Associate searcher at


Alexandre Alaphilippe & Gary Machado

Co-founders of

Author :