Last Friday, Quartz published a short article tantalizingly titled "The world's reaction to Brexit, in emoji." (Brexit is the colloquial shorthand for Britain's vote in favor of exiting the European Union.) While the piece was mostly fluff, it does raise a fascinating question: Can emojis teach us anything about situations like Brexit?
With Twitter reporting that 110 billion emojis have been tweeted since 2014, the idea isn't as fanciful as it might seem. Emojis provide a rich and unexplored prism for understanding the evolution of language and culture.
Inspired by the Quartz headline, this past weekend, I downloaded a sample of 100,000 publicly available tweets using the Twitter API. I pulled data for five Brexit-related hashtags, selected to be representative of both sides of the debate: #NotMyVote, #VoteRemain, #EURef, #Brexit, and #VoteLeave. All tweets were from Friday, June 2, the day after the election, and were in English. After removing retweets, I was left with 23,989 unique tweets (I call this "the Brexit dataset"). Of these tweets, 6.3 percent (1,505) contained at least one emoji. Yay!
First, I found the 10 most commonly used emojis in the dataset (Figure 1). Face with tears of joy and the British flag both crushed it here, appearing, respectively, 11.9 and 10.4 times for every 1,000 tweets. The other emojis are a good mix of both celebratory emojis as well as more melancholy ones.
But are these emojis actually reflective of Brexit, or are some of them just generally popular emojis? Luckily, emojitracker has been tracking every emoji publicly tweeted since mid-2013. So, for each of these ten emojis, I compared how often they appeared in the Brexit dataset with their general popularity.
Five of the ten emojis over-indexed for Brexit (Figure 2). Reasonably, the British flag is over 400 times as likely to be included in a tweet about Brexit than in a random tweet on the platform. Thumbs up and clapping hands both over-indexed at 2.3 times and 3.9 times respectively, while crying face and see-no-evil monkey over-indexed at 1.3 times.
Brexit seems to elicit strong emotions from people on Twitter, on both sides of the spectrum, with slightly more prevalent use of celebratory emojis. This could mean that people disappointed by Brexit are less likely to tweet, or if they do tweet, they might be less likely to use emojis.
I then took a closer look at a few emojis used in context. Even basic emojis can encompass a wide variety of meaning. The British flag, for example, can be used to express nationalism but sometimes it's just a neutral shorthand (Figure 3). Irony is also common, for example, with the thumbs up emoji, which is used both mockingly as well as sincerely (Figure 4).
Can we use data to understand how people on different sides of the Brexit debate use emojis differently? It turns out we can! We know the overall distribution of the five hashtags of interest in our dataset. #Brexit is most prevalent at 29 percent, followed by #EURef at 20 percent, #VoteLeave at 19 percent, #VoteRemain at 17 percent, and #NotMyVote at 15 percent. For a particular emoji, we can look at all tweets that use both that emoji and one of these five hashtags and compare the hashtag distributions to the one above.
For example, we find that 34 percent of tweets with the British flag emoji also include the #VoteLeave hashtag. This is significantly higher than the 19 percent prevalence of #VoteLeave in the overall dataset so we say that #VoteLeave over-indexes with the British flag. The celebratory emojis also over-index with #VoteLeave, while the crying, sad face, and praying emojis over-index on #VoteRemain (Figure 5).
Below, I've included a heat map summarizing the co-occurrence of the five hashtags with the top 10 emojis (Figure 6). It's fascinating to see how distinct the celebratory and sad emojis are in terms of the political sentiments they're used to magnify. It's almost as if the emojis show us #VoteLeave clapping and cheering, while #VoteRemain is crying, distraught, and heartbroken. (Curiously, none of the emojis over-index on #Brexit, suggesting that it may be more of a identifying hashtag for the situation rather than one used by partisans who are emotionally invested in the debate.)
As emojis become more ubiquitous, we should remain open and curious about what they reveal. Does the meaning of emojis change over time? How do we use emojis to talk about war, or DJ Khaled, or our search for love? Will two different people ever see the same emoji the same way?
The Brexit dataset shows us that emojis, far from being trivial objects used ad hoc, have the ability to reflect our emotional states back at us.
Perhaps what makes emojis so exciting is their limitless potential for nuance. Robert Greene wrote that "Paradox is seductive because it plays with meaning. We are secretly oppressed by the rationality in our lives." He might as well have been talking about emojis. Emojis allow us to play with meaning, to inject varying combinations of emphasis, irony, and humor in our communication. While we may never fully understand what they mean, treating emojis as data can help us gradually chip away at the mystery, and in doing so, come to better understand ourselves.
Acknowledgements: I'm indebted to earlier work by Lauren Ancona and Chris Tufts on measuring emoji prevalence among Twitter users in Philadelphia, Jessica Peterka-Bonetta's mapping of emojis from UTF-8 to their R encodings, and of course to Matthew Rothenberg's emojitracker.
Hamdan Azhar is a writer and data scientist based in New York. His writings on politics, culture, and technology have appeared in the Washington Post, the Christian Science Monitor, and the Huffington Post. Read more of his work at hamdanazhar.com.