Can a simple emoji make online toxicity worse? Emojis have become a big part of how we communicate online, helping us express everything from kindness and support to negativity and even harmful behavior. It’s pretty easy to see how emojis contribute to positive interactions, but things get trickier when it comes to their role in spreading and sustaining toxic behavior. Because emojis can be interpreted in so many ways, understanding how they fuel negativity isn’t always straightforward.
In a recently published study, “I 👍Your Hate: Emojis as infrastructural platform violence on Telegram,” researchers Esteban Morales (University of Groningen), Jaigris Hodson (Royal Roads University), Anatoliy Gruzd (Toronto Metropolitan University), & Philip Mai (Toronto Metropolitan University) took a closer look at how emojis are used alongside toxic speech—specifically as tools that can reinforce, spread, and even normalize harmful behavior on digital platforms.
Methodology
Using Communalytic, a computational social science research tool, the researchers gathered 98,729 publicly accessible posts—along with the associated 235,718 emoji reaction counts—from the Chismes Frescos Medellín (Fresh Gossip Medellín) Telegram group. To assess toxicity, they used Detoxify, a machine-learning classifier available in Communalytic’s Civility Module. The Civility Module analyzed every post in the dataset by assigning a toxicity score between 0 and 1 (with 1 indicating a high likelihood of being perceived as toxic). A threshold of 0.7 was used for categorization—posts with a score of 0.7 or higher were classified as toxic (3,155 posts), while those with a score below 0.7 were deemed non-toxic (95,574 posts). A toxicity score of 0.7 suggests that 7 out of 10 people would likely consider the post toxic.
Results
Research Question 1: How are emojis used differently on toxic vs. non-toxic posts?
We relied on descriptive statistics and t-tests to analyze the 235,718 emoji-based reactions. Specifically, we examined their distribution across posts categorized as toxic and non-toxic, and compared the mean number of reactions between these two groups.
Findings
Our results indicate that the average number of emoji reactions per post is greater for toxic posts compared to non-toxic posts. More interestingly, positive emoji reactions (👍,😁,👏,❤️) are more likely to occur on toxic posts than on non-toxic posts (see the figure below).
Research Question 2: How do group members use emoji-based reactions on toxic comments?
To address our second question, we conducted a thematic analysis to identify themes and narratives in the toxic posts that elicited the most emoji reactions.
Findings
While some positive reactions signaled agreement with toxic language attacking others, they were also used in response to posts advocating violent measures.
For example, one post urging the killing and torture of gender-based offenders (“This should be the end for abusers, rapists and femicides“) received multiple positive emoji responses.
Similarly, xenophobic posts—such as discussions on how to handle the surge of Venezuelan migrants—also attracted numerous positive reactions.
This suggests that positive reactions can indicate both endorsement of aggressive rhetoric and support for violent calls to action.
Discussion
- Through the lens of infrastructural platform violence, emojis act as affordances that enable the normalization of harm across everyday uses of digital platforms.
- Our findings show that toxic posts most commonly receive the most positive emojis, suggesting that those who post violent content might be receiving positive reinforcement through the use of emojis for posting the most negative content
- Thus, we need to consider how emojis can escalate already negative and antisocial behaviour further, providing fuel for the fires of online violence via a demonstration of community support for it
If you like to learn more about this new paper, check out the full paper here.
Citation: Morales, E., Hodson, J., Gruzd, A., & Mai, P. (2025). I👍 your Hate: Emojis as Infrastructural Platform Violence on Telegram. https://hdl.handle.net/10125/109124