Jun 27, 2018 7:00 AM

Academics Gathered to Share Emoji Research, and It Was 🔥

Linguists and data scientists see a new way to study language and communication in our little digital ideograms.

Two years ago, Sanjaya Wijeratne—a computer science PhD student at Wright State University—noticed something odd in his research. He was studying the communication of gang members on Twitter. Among the grandstanding about drugs and money, he found gang members repeatedly dropping the ⛽ emoji in their tweets.

Wijeratne had been working on separate research relating to word-sense disambiguation, a field of computational linguistics that looks at how words take on multiple meanings. The use of ⛽ jumped out as a brand new problem. “They were using the gas pump emoji to refer to marijuana,” says Wijeratne. “As soon as I saw this new meaning associated with the emoji, I thought, what about emoji-sense disambiguation?”

That moment caused Wijeratne to redirect his PhD research toward emoji. This week, he put together the first interdisciplinary academic conference on emoji in research.

At Stanford University this week, a collection of linguists, data scientists, computer researchers, and emoji enthusiasts gathered for the International Workshop on Emoji Understanding and Applications in Social Media, itself a smaller piece of the AAAI Conference on Web and Social Media. They brought with them research on how emoji are changing the way we communicate online, how gender and political affiliation are reproduced online through emoji, and the challenges emoji pose for natural-language processing in computers. The assembled academics also debated basic questions about the nature of emoji: Like, if emoji is something akin to a language, why can’t anyone agree on what individual emoji mean?

Emoji, which have grown from an original set of 176 characters to a collection of over 3,000 unique icons, present both opportunities and challenges to the academics who study them. Most agree that the icons are not quite a language—the emoji vocabulary is made up almost entirely of nouns, and there’s no real grammar or syntax to govern their use—but their influence on internet communication is massive. By 2015, half of all comments on Instagram included an emoji. On Messenger, Facebook’s messaging app, over 5 billion emoji are sent and received every day. From an academic point of view, that presents a wealth of data to understand communication, behavior, and language online.

But the academic research on emoji has, until recently, been limited. Earlier gatherings like EmojiCon, which will have its second conference this summer, have brought emoji conversations to the mainstream. But that event—a “celebration of all things emoji”—courts a popular audience, and feels less like a formal conference and more like a party made for Instagram. This week's Workshop on Emoji Understanding, on the other hand, brought the focus squarely back into academia. The day-long event included a series of paper presentations that privileged data sets and citations over emoji-shaped balloons, and asked more questions than it could answer.

Papers presented at the conference highlighted emoji as markers of solidarity during crisis (think: “Je suis Paris 🙏🇫🇷”) or as ways to understand differences across gender or political ideologies (women use emoji more than men, but conservative men use way fewer emoji than liberal men). Others discussed the potential to decode emoji with machine learning, and the difficulties in teaching computers to recognize the multiple meanings of emoji in natural-language processing. A panel discussion raised questions about the way the emoji lexicon is developed, as well as the ways emoji can be misinterpreted across cultures. (The 👌 does not mean the same thing in English as it does in American Sign Language, nor does it mean the same thing to white supremacists.)

Tyler Schnoebelen, who gave the keynote speech on Monday, says conversations about emoji have been too often painted with a broad brush. There’s the utopian vision: emoji as a "universal language," the great democratizer and harbinger of communication across class, culture, and geography. And then there’s the doomsday vision: emoji as the destruction of language, a political tool, a new way to send violent threats. The nuance often gets lost in between. We have hardly any research to tell us who uses emoji, when, why, and how that use has changed over time. We know even less about what emoji can reveal in disaster scenarios, campaigns, or educational settings; even linguists, who have looked at emoticons and other internet-born languages for decades, don't have a consensus on what emoji mean for the future of language.

Now, researchers are beginning to turn more seriously toward those research questions. On Monday, linguist Gretchen Mcculloch presented a theory of emoji as beat gestures—the equivalent of gesticulating to add emphasis—rather than a language in themselves. "Letters let us write words, emoji let us write gestures," she says. Eric Goldman, a legal scholar at Santa Clara University's School of Law, discussed a forthcoming paper on emoji and the law, which highlights the potential for emoji to create misunderstanding in legal contexts—including high profile cases, like the Silk Road case.

Other scholars are looking for ways to incorporate emoji into preexisting research. “We do a lot of social media research: depression on social media, harassment on social media, the opioid crisis on social media,” says Amit Sheth, a computer scientist at Wright State University and co-organizer of the conference. “In all of those problems, we also see significant use of emoji. If you were to only study the text, you’d be missing out on a lot of information.”

As the conference wrapped up, researchers from institutions in the United States, Spain, India, and Germany shook hands and traded email addresses. That, Wijeratne says, is the point of the event: not to answer questions about the role of emoji in our world, but to connect researchers from around the world and spark ideas for future studies.

Correction on 6/27/18 at 11:45 a.m. ET: EmojiCon will have its second conference this year, not its third.