Abstract

The following Paper is an analysis of the twitter behavior of four well known German politicians, three of which were candidates for the CDU party leader. To do so, the timelines of the past 20 months of each politician were extracted and analyzed. This was done in three steps. First the the general data and information is described. After that the content of each tweet is analyzed and put into a comparable form. Last there will be a short look into the followers of each politician. In the end We will have the answer to the question, whether the twitter bahivior of the candidates changed after declaring their candidacy, and how it changed.

1. Motivation

Similar to the social and economic life, the new Corona virus has also changed the political landscape. Whilst the CDU, Germany’s biggest political party, would have gotten 26% of the votes in March 2020, that number rose to 36% as of January 2021 1. With a vote share last seen this high in 2017, it is quite likely for the next leader of the CDU to also become Germany’s next chancellor which is reason enough to have a closer look into this subject.

Social media has constantly gained importance in the political world. In 2017, 32% of Germans used Twitter and 27% claimed to get their information - including political information - from social media 2. Therefore, it is not a surprising phenomenon that politicians use twitter to promote themselves, their agenda or just communicate with their voters. Especially the AfD, a right-wing populist and controversial party, has discovered Twitter as a cheap marketing tool, having more tweets and far more likes and retweets than any other political party 3. The field of political marketing and communication is shifting and social media has become an inevitable part of it. Buzz marketing techniques are taking over which is shown in a study conducted by Safiullah et al.: They found a correlation between the social media buzz of political parties and the seats won in the 2014 general election of India 4. Based on this, the goal of this paper is to analyze the Twitter accounts of the three CDU party leader candidates: Norbert Röttgen, Armin Laschet and Friedrich Merz. Eventually, Markus Söder is also taken into account, due to the recent talk about him being a possible chancellor candidate. He is also taken in as a comparison to spot possible differences between his Twitter behavior and the behavior of the actual candidates. In addition, close attention shall be paid to any changes in their Twitter behavior since declaring their candidacy.

2. General Analysis

In order to get a better understanding of the four politicians, a look at their Twitter account information and the source of their tweets shall be taken. Their Twitter activity shall be monitored, analyzing how much they tweet and finding out what time of the day they chose to send their tweets. Lastly, there will be an analysis of the accounts most retweeted by the politicians as well as the reactions to the tweets the politicians post. The aim of this part is to gain an overview over the politician’s Twitter accounts but also to make first comparisons between before and after candidacy data. The data set used here contains 20 months of Twitter data. Ten months before the candidacy race and ten months after. In total, the data consists of around 10,000 tweets and was extracted using the retweet package5. For ease of reference, the time before candidacy will be referred to as 2019 and the time after candidacy as 2020. Having said that, the analyzed time periods have also mainly taken place in these years.

2.1 Twitter Account Information

First of all, a closer look at the general Twitter data of the politicians shall be taken. What stands out is that Söder and Laschet have had a Twitter account since 2012, while Merz and Röttgen have joined Twitter more recently in 2018. Apart from that, the attributes are rather inconsistently spread. With 223,777 Followers Söder has the most which is most likely explained by the fact that he is the only politician in the list being a Minister-President and a Party Leader. Laschet has the most Tweets (15,597) but less Followers (92,644), taking into account that he is the Minister-President of the most populated German federal state. With Röttgen being a new face in federal politics, it is rather unsurprising that he has the fewest followers (46,374). In comparison, Merz has more followers (123,852) but the least tweets (823). This data provided reveals that there are big differences between the politicians in regards to their Twitter data. It also shows that extracting the timelines over a longer period of time is rather unnecessary for these subjects, due to their moderate tweeting behavior.

Name Location Followers Friends Tweets & Retweets Account Created
Markus Söder Nürnberg 223777 277 3253 2012-08-31 12:38:50
Friedrich Merz Deutschland 123852 464 823 2018-11-05 10:43:23
Armin Laschet Aachen / Düsseldorf 92644 1178 15597 2012-09-20 13:13:25
Norbert Röttgen Berlin / Rhein-Sieg-Kreis II 46374 1961 2904 2018-09-13 08:50:34

2.2 Source of Tweets

The most common source for the Tweets of Röttgen, Laschet and Söder is the Twitter app. The only exception is Merz who uses TweetDeck, a collaboration tool for managing multiple Twitter accounts. The politicians are not using any Twitter bots which means that they do not use automated tweets. Still, their tweets could be written and posted by someone else.

Politician Source Number of Tweets
Armin Laschet Twitter for iPhone 3175
Friedrich Merz TweetDeck 396
Markus Söder Twitter for iPhone 3028
Norbert Röttgen Twitter Web App 1548

2.3 Tweets per Month

Röttgen announced his candidacy on the 18th of January 6. He was the last candidate to enter the political ring, therefore this date will be taken for an before & after comparison. The following graph shows, 10 months of twitter before the 10th of January and the 10 months after.

As we can see, every politician except for Laschet has significantly increased their twitter activity in the last 10 months. In the 10 months after entering the race, Laschet actually decreased his tweet frequency by 36%. The biggest increase can be seen by Röttgen with 240%, followed by Söder with a 119% and Merz with a 45% increase in tweets.

Politician Tweets 2019 Tweets 2020
Armin Laschet 1197 771
Friedrich Merz 289 419
Markus Söder 353 773
Norbert Röttgen 578 1961

The Results however are biased by the corona pandemic. You can see a peak of tweets in March, 2020, the same month the WHO declared corona a pandemic7 and Germany had its first lookdown8. The two Minister-Presidents, Söder and Laschet, sent out the most tweets during that month, mainly to spread the rapidly changing information as fast as possible to their Region. The clearest trend can be seen in the twitter activity of Röttgen. In the last 6 Months he has gradually increased his tweets per month. An increase which probably is necessary due to him being the only one of the four not in the top ten of the most important German politicians 9. On the contrary, Laschet’s twitter activity does not indicate any special behavior in terms of tweet quantity. In fact by the looks, it seems as if 2020 had been a rather unspectacular year, which is of course not the case. As mentioned before, both Söder and Merz had a significant increase in their twitter activity. It is however not completely out of proportion.

2.4 Most Frequent Twitter Times

To complete the bigger picture, we also want to have a look at the times the politicians sent out the most tweets. This is done by separating the timeline data into 2019 and 2020, extracting the hour of the time stamps and summing these up using the lubridate package10.

We can see, that before the candidacy in February 2020 Merz, Laschet and Söder have similar tweeting times. They tend to tweet less in the morning, reach their peak at noon and slow down their twitter activity reaching the evening. The only bigger exception here is Röttgen. He has his peak time at 8 o’clock in the morning, and then gradually decreases his twitter frequency until he reaches the evening. He also has the biggest changes in his twitter activity. While running for party leader Röttgen gained a second tweeting activity peak at 8 o’clock in the evening. In fact all politicians developed a little 8pm activity peak in the last 10 month. However in direct comparison the changes of the twitter activity of Laschet, Söder and Merz are not as big. One can only assume, that this change is connected to the corona pandemic. A possible explanation is, that the conferences for new corona rules are usually held midday. This pushes the twitter times of the minister presidents into a later time of the same day. With news articles getting published with a few hours of delay, the information density also reaches its peak in the evening, giving more twitter topics for all politicians.

2.5 Retweetrate

In this section we want to have a look at the retweet rate and the reweet sources of each politician. We can see, that on average the retweet rate lies at around 20%. Lashet is the only exception in this list. His retweet rate is at 70%. This is interesting, because in reverse this means that only 30% of his tweets are originally written by him or his delegates. Given the fact that he has already slowed down his twitter activity in 2020, this diminishes his activities even more.

Friedrich Merz Markus Söder Norbert Röttgen Armin Laschet
0.163 0.189 0.235 0.700

2.5.1 Retweets of Armin Laschet

By taking a look at Laschet’s retweeted accounts we can get an explanation for his high retweet rate. The most retweeted account by far is “Staatskanzlei NRW”, basically his own employees. Their main topics are North Rhine-Westphalia and Armin Laschet. So he is retweeting his own press office, which makes sense. This of course increases the chances, that he himself is tweeting on his account and is only including his press office on his private twitter by retweeting them. His retweet sources have not changed dramatically compared to 2019. The main retweeted account is still “Staatskanzlei NRW”, followed by very few almost randomly retweeted accounts connected to politics or news. In 2020 he also retweeted his supporter Jens Spahn, but not on a high scale.

2.5.2 Retweets of Friedrich Merz

The account most retweeted by Merz is similar to Lashet, his press officer Armin Peter. The difference here is the smaller scale and that the press officer is tweeting from his private account. Apart from that he only retweeted CDU accounts, fellow CDU politicians and initiatives with ties to and founded by the CDU. Although the retweet accounts slightly changed, the agenda of the accounts didn’t. Merz did not have a notable change in his retweets from 2019 to 2020. His most retweeted accounts have been and are still very political.

2.5.3 Retweets of Norbert Röttgen

Röttgen does not have an account he retweets far more than other accounts. In 2019 his most retweeted accounts were a plitical talkshow, national german radio and an independent institute. This slightly shifted in 2020 where “CDU Deutschland” also made it into the top three most retweeted accounts, showing that his retweets had a more political touch in 2020.

2.5.4 Retweets of Markus Söder

Leaving the quantity aside, Söder’s retweet behavior did not change much from 2019 to 2020. His top two are allways “BR24” and “CSU”. Only the political talkshow on his number three changed from “ANNE WILL talkshow” to “maybrit illner”. Given that he is not officially running for anything at the moment, one could argue that his retweeted accounts reflect exactly that.

3 Tweet Contentanalysis

In this part of the paper we want to have a closer look at the content of the tweets. To analyze the tweets we will look at the hashtags, that are attached to each tweet. After that we will analyze the content of the tweets followed by an analysis of the followers interactions with the tweets.

3.1 Twitter Hashtags

To get a first oversight, hashtags are a good instrument. They give a first impression on the content delivered in the tweets. Hashtag data can easily be extracted and there is not much cleaning needed to get a usable result. Of course hashtags such as “corona” and “Covid-19” are omnipresent in the 2020 twitter timelines. They were the most or a least second most used hashtags by every politician. Having that said we will not dig further into the unsurprising fact, that corona was a trending topic in 2020. We will have a look at hashtag changes apart from corona.

3.1.1 Hashtags of Armin Laschet

In 2019 the most used hashtag of Laschet was “NRW”, the state he is governing. It is followed by far less used hashtags such as “NRWKoalition” and “Kuturbevollmächtigter”. “Merkel” is also among the top hashtags in 2020, while “NRW” is still the most used hashtag. the new appearance of the “Laschet” hashtag has its origin in posts that promoted him and he therefore retweeted.

3.1.2 Hashtags of Friedrich Merz

Merz made quite a turn with his hashtags. While mentioning only his rivaling parties “SPD” and “AfD” but not his own party in 2019 this changed in 2020 where his most used hashtag became “CDU” not mentioning any other party in his top five hashtags. His top hashtag in 2019 was “merzrede”, a hashtag he used when posting his own speaches. Overall we can see a clear topic shift from other parties back to the CDU.

3.1.3 Hashtags of Norbert Röttgen

Beeing the head of the foreign committee Röttgens top hashtags in 2019 circled around Europe, China and Brexit. His most used hashtag was “5G”. In 2020 he kept a part of his hashtags dedicated to foreign topics. His most used hashtag in 2020 is “CDU” and also the hashtag “jetztvoran”, his campaign slogan.

3.1.4 Hashtags of Markus Söder

If there had not been a pandemic, Söders hashtags would have probably not changed much. The most used hashtags of both years are “Söder” and “Bayern”. In 2019 “Nürnberg”, “AnneWill” and “CSU” were also frequently used hashtags. In 2020 these hashtags just got replaced by the already mentioned pandemic.

3.2 Tweet Frequency Analysis

After getting a first impression with the help of hashtags, we no proceed to the actual content of the tweets. This was done by extracting the twitter texts and cleaning them from links and punctuation marks. All words were mutated into low caps to avoid redundancies. The result is a wordcloud representing the frequency of words in size.

3.2.1 Wordcloud of Armin Laschet

We can see that Laschet’s most central topic of 2019 was NRW. Keeping his hashtags in mind, this is no surprise. Apart from geographics such as Germany, Cologne and Europe, climate change was also a dominating topic. Words like environmental protect, coal exit and energy transition are words often mentioned in 2019. This is most likely also connected to the fridays for future movement, which had its first international climate strike in on the 15th of March, 2019, and kept growing stronger throughout the entire year11.

Laschets 2020 tweets were dominated by again NRW and of course corona. Coronawise the tweets have a uniting and dramatic character. Words such as Krisis, musst and stay, but also the words teogether, unity, and “nrwkanndas” were mentioned often. We can also see, that climate change has lost its place in the agenda. From a political perspective the only change is a more frequent usage of the word cdu, which has not apeared in 2019’s wordcloud

3.2.2 Wordcloud of Friedrich Merz

The central topics of Merz 2020 were Germany and CDU. The word “must” is also very frequently used, indicating a rather demanding tone even before the pandemic. Politics resembles the core of these tweets. This shows, that after losing the election for party leader against Kramp-Karrenbauer in December 201812, Merz did not leave the political landscape and stayed very involved in sharing his opinion on politics.

Merz’s 2020 tweets kept their political profile. On top of that came words such as “cduvorsitz”, departure and teammerz showing his candidacy very obviously in his twitter feed.

3.2.3 Wordcloud of Norbert Röttgen

As the head of the foreign committee Röttgen’s twitter content circles mainly around foreign politics. His most used words are Brexit, China, USA, Iran, and Europe. Röttgen is the only one of our politicians, who tweets in English and German. This also reflects the wordcloud. CDU is also a word, that has been used very often.

We can see an increase in the usage of the word CDU. Foreign politic is not a central topic anymore it has decreased by a lot. Germany is mentioned a lot more and also words indicating his candidacy such as “cduvorsitz”, teamröttgen and jetztvoran, the last one being his campaign slogan.

To highlight Röttgen’s shift towards tweeting in German, I have extracted his English, as well as his German timeline. We can see, that in 2019, there were weeks in which Röttgen sent of more tweets in English language, than in German. We can see how drastic this changes in the end of 2020. Röttgen has not only changed topics from foreign to domestic, but also his language from English to German.

3.2.4 Wordcloud of Markus Söder

Söder had one favorite topic in 2019: Bavaria. Apart from that there is not really much going on in Söder’s twitter feed. Climate protection is also partially mentioned, apart from that most words are connected to bavarian cities or with little room for interpretation.

2020 gave Söder a reason to twitter outside the box. Corona was Söder’s main topic in 2020.

3.3 Sentiment Analysis

As a least step of our content analysis we will conduct a sentiment analysis. In this analysis the words used in the tweets will be rated on scale from -1 for very negative to 1 for very positive. Any values in between can occur as a rating for a word. For our sentiment analysis a dictionary including 3471 negative and positive words from the University of Leipzig13 and the tm package14 are used.

In the first table we can see the average sentiment rating of the politicians in 2019 and 2020. It shows that on average Laschet’s tweets became slightly more, and Röttgen’s tweets far more positive. In Röttgen’s case he even went from a negative sentiment in 2019 to a positve one in 2020. Söder became slightly more negative as well as Merz. Merz is the only one with a negative sentiment in 2020. These results suggest, that the mood delivered by the politician is more dependent on the type of person than on the job itself. Söder and Laschet both are minister-presidents but had opposite developments in their sentiment.

Politician Sentiment in 2019 Sentiment in 2020
Armin Laschet 20.5402 25.0171
Friedrich Merz -0.2021 -5.3427
Markus Söder 12.0232 4.7741
Norbert Röttgen -10.0528 34.1894

Here we can see a far more detailed version of the sentimens in 2020. We can see the average positive and negative sentiment of each month from February 2020 until January 2020. Overall every politician has had a very continuous pattern over the last 12 months. Corona as well as the candidacy did not seem to have a big impact on the sentiment of our politicians. The only exception here is Röttgen, his tweets became more positive. The text analysis showed, that the content of his tweets differed very strongly in 2020 and 2019. This sentiment however illustrates, that the tone chosen to communicate is not always dependent on topics, but on the politician himself, and has therefore not changed in three out of four cases.

4. Followers

Of course the followers of each politician play a big role when it comes to analyzing their profile. We will do this by first observing the interaction of the followers with the tweets of their politician. This is then followed by looking at the location of the followers and finally we will also look at mutual friends of the politicians. A frequency analysis of the followers’ timelines will not be conducted. Due to twitter limitations, a big sample of these timelines could not be extracted. The frequency analysis of these timelines had no usable output. It is therefore excluded form this paper.

4.1 Followers’ Interaction

Tweets alone are not worth much. Therefore it is also important to look at the reactions to the tweets. In the following graphic you can see the likes and dislikes for the tweets of each candidate.

In order to visualize the data, outliers had be taken out. The data was cut of at 4000 likes and retweets. This excludes 58 Tweets from the data set. All of the tweets exceeding 4000 retweets were retweeted themselves by the politician. Almost all of the the tweets excluded because of to many likes were original tweets. The outliers are on average either strongly emotional or polarizing. Some topics of these tweets are: Corona, Notre Dame, Attacks on Vienna, Lübcke assassination, Brexit, Immigration and Burkas.

On average, the likes and retweets on Laschets tweets have increased. His tweets have the tendency to either be retweeted or get favorites but never both. Merz tweets have only gained favorites. The retweets on his tweets remain low, in fact they have decreased compared to the year before. Söder has the highest increase in favorites, his retweets have also increased. Röttgen has also significantly gained favorites and retweets. He is the only one who has a tendency to have many favorites and retweets at the same time on his tweets

The graphic shows, that especially Röttgen gained a lot of twitter attention and increased his activity on twitter most likely because of his candidacy. The other person who sticks out in terms of activity and attention is Söder. In his case it is most likely because of his role and image during the corona pandemic.

4.2 Followers’ Location

After looking at the interaction of the followers, we also want to have a quick look at their location. Therefore the locations of our in total 60,000 sample followers were collected and arranged descending. These will hopefully help us better understand if a politician has regional or more widespread popularity.

4.2.1 Location of Armin Laschet Followers

While writing this paper Laschet was elected the new party leader of the CDU. This gave him a real twitter follower boost. He climbed from 92,000 followers in early January to 110,000 on the 20th of January. Reason enough to have another look at these followers. In the left graphic we can see Laschet’s followers before his election. We can see, that six out of ten cities in this list are in NRW. The rest are big German cities such as Berlin and Munich. This changed very dramatically after Laschet’s election. The top cities of his followers became a lot more international, suddenly having cities such as London, Brussels, Paris and Istanbul on his list. A possible explanation for this shift is the received international attention especially from twitter users involved in politics. Politicians are far more likely than the average twitter user to share their location. Laschet has clearly gotten more international followers, but these new gained twitter followers might also share their location more than other followers, due to their possible involement in politics. And again this sample only contains around 10% of Laschet’s followers. But a tendency to more international twitter followers can not be denied.

4.2.2 Location of Friedrich Merz Followers

Merz’s most common follower locations are proportionally distributed over the most populated cities in Germany.

4.2.3 Location of Norbert Röttgen Followers

The followers of Röttgen have similar location as Merz, probably for the same reasons.

4.2.4 Location of Markus Söder Followers

Similar to Laschet before his election, Söder also has more followers in cities, that are located in his state. Four out of the top ten cites are in Bavaria.

4.3 Friends Matrix

A way to better understand the relation between twitter accounts is a friends matrix. This matrix shows us how many common friends the politicians have. The accounts you follow on twitter are considered your friends. As we can see below, Söder has the most common ground with his fellow politicians, when it comes to following the same accounts. The three candidates Merz, Laschet and Röttgen actually do not many common accounts among each other. This is quite interesting, because one would think that the three candidates would have more similar sources on twitter they rely on.

5. Conclusion

In this paper we analyzed the Twitter activity of Armin Laschet, Friedrich Merz and Norbert Röttgen, the three candidates wanting to become the new CDU party leader. As well as Markus Söder, a name often mentioned in talks about potential chancellor candidates. Each politician had their own, but very constant sentiment level. The retweets and likes on their post increased for every politician in 2020. The Location of their followers was connected to their residency, with for example Söder having more followers from bavarian cities than Röttgen. A FriendsMatrix showed that Söder has the more mutual friends with the candidates, than the candidates among themselves.

Laschet has reduced his twitter activity in the year after declaring his candidacy. He has a very high retweet rate of around 70%. The main account he retweets is his own press office. The content analysis of Laschet’s tweets showed, that his main topic always is NRW. In 2020 his main topic was corona. By only looking at Laschets most used words, one would not notice that he was running for CDU leader. The sentiments of his tweets are continuously positive, with a little increase in 2020. After winning the election Laschet’s followers increased by almost 20%. On top of that, a large portion of these new followers come from European cities outside of Germany. Merz slightly increased his tweets in 2020. His most retweeted accounts in both years are all connected to the CDU. His twitter content has been very political before his candidacy and has gotten even more political after. He shifted his focus from other political parties to his own after running for election. Corona did not play a big role in his tweets of 2020. Merz is the politician with the lowest and at the same time the only one with a negative sentiment score. He is using a more negative tone than his competitors. Röttgen increased his twitter activity by 240% after declaring his candidacy. His content shifted from foreign to domestic topics. He was the only candidate also tweeting in English. This also changed in 2020, where his German tweets outnumbered his English ones by far. Corona has only played a marginal role in his feed. His sentiment score has become far more positive in 2020. Söder has also increased his twitter activity in 2020. His main topic in 2019 was Bavaria, this topic got competition in 2020 by Corona. Apart from that nothing much changed. His sentiment score is slightly over Zero, indicating a neutral tone in his tweets.

This paper was able to show, that from the point of declaring their candidacy the twitter behavior of Merz and Röttgen changed in a very noticeable way. It also shows that the twitter activity of Laschet, the winner of this race, did almost not change at all.

Inhaltsverzeichnis

  1. Wahlrecht.de. (2021). Sonntagsfrage – Forsa (Wahlumfragen Zur Bundestagswahl). [online] Available at: https://www.wahlrecht.de/umfragen/forsa.htm [Accessed 21 January 2021].↩︎

  2. Hrsg. pwc (2018) “Vertrauen in Medien - 2018”. [online] Available at: https://www.pwc.de/de/technologie-medien-und-telekommunikation/pwc-studie-vertrauen-in-medien-2018.pdf [Accessed 21 January 2021].↩︎

  3. Serrano, J. C. M., Shahrezaye, M., Papakyriakopoulos, O., & Hegelich, S. (2019). The Rise of Germany’s AfD: A Social Media Analysis. In Proceedings of the 10th International Conference on Social Media and Society (pp. 214-223).↩︎

  4. Safiullah, M., Pathak, P., Singh, S., & Anshul, A. (2017). Social media as an upcoming tool for political marketing effectiveness. Asia Pacific Management Review, 22(1), 10-15.↩︎

  5. Kearney, MW. (2019). “rtweet: Collecting and analyzing Twitter data.” Journal of Open Source Software, 4(42), 1829. doi: 10.21105/joss.01829, R package version 0.7.0, https://joss.theoj.org/papers/10.21105/joss.01829.↩︎

  6. Handelsblatt.com. (2020). Analyse: Röttgen, Der Partycrasher Der CDU. [online] Available at: https://www.handelsblatt.com/politik/deutschland/analyse-roettgen-der-partycrasher-der-cdu/25557272.html?ticket=ST-4640543-XfLyXmUmll7c7wS7SNto-ap1 [Accessed 21 January 2021].↩︎

  7. Euro.who.int. (2020). WHO Erklärt COVID-19-Ausbruch Zur Pandemie. [online] Available at: https://www.euro.who.int/de/health-topics/health-emergencies/coronavirus-covid-19/news/news/2020/3/who-announces-covid-19-outbreak-a-pandemic [Accessed 21 January 2021].↩︎

  8. Mdr.de. (2021). Jahresrückblick 2020 Zum Coronavirus | MDR.DE. [online] Available at: https://www.mdr.de/nachrichten/politik/corona-chronik-chronologie-coronavirus-100.html#sprung2 [Accessed 21 January 2021].↩︎

  9. Forschungsgruppe.de. (2021). Politbarometer Januar | 2021. [online] Available at: https://www.forschungsgruppe.de/Aktuelles/Politbarometer/ [Accessed 21 January 2021].↩︎

  10. Grolemund, G., Wickham, H. (2011). “Dates and Times Made Easy with lubridate.” Journal of Statistical Software, 40(3), 1–25. https://www.jstatsoft.org/v40/i03/.↩︎

  11. Wetzel, J., (2019). München: Ein Jahr Fridays For Future. [online] Süddeutsche.de. Available at: https://www.sueddeutsche.de/muenchen/klimastreik-muenchen-fridays-for-future-1.4718739 [Accessed 21 January 2021].↩︎

  12. Braun, S., (2018). Warum Kramp-Karrenbauer Gewann Und Merz Verlor. [online] Süddeutsche.de. Available at: https://www.sueddeutsche.de/politik/cdu-parteitag-kramp-karrenbauer-merz-1.4245059 [Accessed 21 January 2021].↩︎

  13. Remus, R., Quasthoff U. & Heyer, G.: SentiWS - a Publicly Available German-language Resource for Sentiment Analysis. Proceedings of the 7th International Language Ressources and Evaluation (LREC’10), 2010↩︎

  14. Feinerer, I., Hornik, K. (2020). tm: Text Mining Package. R package version 0.7-8, https://CRAN.R-project.org/package=tm.↩︎