The background

Politicians ONLY

Our story starts with the Quotebank dataset, a collection of quotes gathered from English news articles published throughout the years. First, we filter the dataset by the profession of the predicted speaker (must be a politician, duh). After, we select only the quotes whose speaker we can associate with a party, as it's not enough to be a politician - you need to represent a party at the time you were quoted.

Starting with 115.584.257 quotes, we end up with 9.801.277 quotes, each attributed to a politician and a party, after enriching our data with the mighty Wikidata.

It's all about the words

Words, words, words. We use words everyday to express our needs, discuss abstract concepts, and, of course, pointlessly argue about politics. Politicians are no different, but their speeches often have a deeper goal: to be relatable to their possible voters. They have advisors that help them craft a persuasive speech, and analyze the strategy of the opposition.

Therefore, we ask ourselves: what is common and what is unique to each party? What sort of words do they employ to convey their message?

I know words. I have the best words.
— Donald Trump

We take a look at the two most prominent parties in the U.S., the Democrats and the Republicans, to find out whether the words they use the most are just as different as their ideologies.



Most common nouns & verbs by party

Rank Nouns Proper Nouns Verbs
Democrats Republicans Democrats Republicans Democrats Republicans
1 people people Trump Trump will will
2 time time New President can going
3 state state America United going think
4 country country President States think can
5 way way Congress America would would

Politicians can!. This inspires trust and creates the expectation of a different future under their leadership. They think and they would, but most importantly, they will. Trump seems to be on everyone's lips, no matter their affiliation. However, probably the most interesting finding is that both parties share the 5 most used nouns. Make it clear that you're addressing the people, remind them that it's time for change, and you will find the best way to bring it into their state and country.

What's different?

Even though the top 5 words of both parties are similar, we expand our search to the top 50 most used words, to find what makes the speech of the Democrats and the Republicans different. We focus on the nouns that are used by a party, but not by the other.

The Republicans do not mention women, community, and school as often as the Democrats. Instead, they talk about security, deal, and budget. The Republicans discuss more about Obamacare than Barack Obama's party does. Among the most used proper nouns, they bring up places that diverge from a cultural perspective.
Could these differences in the most common words be representative of the interests of the parties? We put on our detective hats and dig on.

Parties & States

Among the most commonly used nouns by party, there are states and countries - probably as a way to connect with their residents (if not to discuss current conflicts). In the United States, Washington is the most talked about state for both parties, which can be explained by its relevance to politics. Next in line, the Republicans have a strong preference for Florida, and Texas, whereas the Democrats talk about New York and California.

Republicans' mentions

Democrats' mentions

Can we guess the election results based on the states parties mention?

We've seen that parties have a preference for the states they mention. We take it one step further and we try to figure out how the 2020 election played out based on the quotes by party.

A bit of tech talk: We first count the number of mentions of each state by party in 2020. However, simply comparing the number of quotes would not be a fair comparison, as some parties might be more vocal than others. Therefore, we standardize the counts for each party and we compare the resulting values. If the Republicans have a higher value, we deem them as winners for that state.

Election winners based on 2020 quotes

Actual 2020 election results

~Whoaa~ It seems like we can use the difference in the number of mentions per state to predict with a pretty high accuracy how states will vote. And it's not just about the 2020 quotes - even using past data on quotes would result in a similar map. Multiple explanations emerge: politicians may talk about the states which already have a preference for them to excite the electorate, they may talk more about states whom they know are interested in what they are saying, or, even more bold, they influence votes based on which states they mention the most.

Okay, okay
but what are the quotes about?

So far we've only looked at words, not concepts. We invoke the power of BERT to create meaningful encodings for our quotes. We create clusters out of quotes with similar encodings, and then we use our collective knowledge™ to label the clusters based on their most representative words.

2015 2019

Between 2015 and 2019, the world has changed a lot. In 2019, more quotes discuss international topics, related to Indian politics, Eastern Asia, and Oceania - possibly due to a larger media coverage. On the other hand, some topics are always on our mind, such as education, women, and transportation. As tensions about the United Kingdom leaving the European Union rise, the Brexit topic gains ground.

2019 in the United States

We all know the war between Republicans and Democrats. However, based on the topics they discuss, they do not seem to be that much different. Among the most popular topics, they both talk about women, education and healthcare, which are signs of a healthy nation, that can focus on non-primary needs.

Much like the most common words, the Republicans' topics show us once again their obsession with East Asia (a topic defined by North Korea and China). However, East Asia is not the only part of the world they talk about: Russia, South America and Turkey/Syria are common topics too. Notably, their quotes are proof that Trump, the most vocal of the Republicans, has a real problem with Mexican immigrants, as the border topic is present in 2.52% of their quotes.

Alternatively, the Democrats are more interested in one of the hottest topics (and for good reason) of the 21st century: gender equality. Unlike the Republicans, they are invested in climate change and law enforcement.

Parties' topics over time

We take a closer look at what the main U.S. parties talk about and how these topics change with years.


Democrats' trends

Unfortunately, guns seem to be one of the main topics in the last 5 years. More than that, the spikes we observe seem to follow the main mass shootings in the U.S.:

  • San Bernardino attack in December 2015
  • Orlando nightclub shooting in June 2016
  • Las Vegas shooting in October 2017
  • Stoneman Douglas High School shooting in February 2018
  • El Paso Walmart shooting in August 2019

Every time there is a shooting, there seems to be an increase in the number of quotes. Two hypotheses can explain this trend. First, it might be possible that the media is more prone to pick quotes related to violence. Second, shootings might influence the Democrats to talk more about gun laws and possible restrictions to limit the chances of such events happening in the future.

Political events bring buzz around voting. Campaigns are all about chattery, so it's not surprising that we see a spike during the following periods:

  • Elections for the House of Representatives and for the President in November 2016
  • Midterm elections in November 2018



Republicans' trends

With years, the Republican party seems to be more interested in international relations, particularly with China, in detriment of other topics such as education, which have a decreasing trend. It's interesting to notice that the biggest spike in the education trends was during the Stoneman Douglas High School shooting, a subject that both parties thoroughly covered.

The tax topic was the hottest in November 2017, when The Tax Cut and Jobs Act was introduced in the House of Representatives.

All in all, there is no deny that politicians discuss the main events happening in the world, as they are of interest to the electorate, and the topic trends suggest this pattern.

All different, yet connected

The relations between parties around the world, through their quotes

The graph portrays the links between parties based on the topics they talk about. There is a clear connection between parties that live in the same geographical location, indicating that they share common interests, specific to their country and the needs of its people. It seems that the geography is even more relevant than the ideology of a party when considering the topics they talk about.