Linguistic Experiments And Emoji Interoperability
The Emoji Standards and Research Working Group (ESR) is responsible for adding new emoji to the Unicode Standard. But standardisation does not end with addition. We often need to align vendors on design that communicates, without simply dictating style and aesthetics.
The Emoji Standards and Research Working Group (ESR) is responsible for adding new emoji to the Unicode Standard. However, standardisation only begins with encoding. Design played a minor role in the first decade of modern emoji expansion: the consortium established the codepoints and left implementation to vendor discretion. This resulted in a fractured landscape where vendors lacked the context needed to ensure interoperability. For those first ten years, adding new characters was seen as the primary way to improve the standard, bringing us close to a massive library of over 4,000 characters today.
As the ecosystem has matured, so has the process to improve the emoji program. Design and research now play more active roles in the encoding process to prevent cross-platform miscommunication and improve the practical execution of emoji across the internet. Further, adding new characters is no longer enough: refining existing emoji and reducing platform fragmentation is now critical if emoji are to remain a thriving, useful digital primitive.
We do this through a mix of expert volunteers and the use of large scale human linguistic experiments. This session will present methods, results and outcomes from one long-running research area: Interoperability.
Interoperability matters because emoji most commonly express tone or emotional intent. Emoji are the standardised means by which people can not only communicate, but also actively avoid miscommunication. The ESR’s Interoperability initiative is ensuring that any emoji, regardless of which platform it is sent from or to, or which vendor designed it, will be interpreted in the same way by both reader and author.
In this 30-minute presentation, we will share data gathered from perception studies involving over 2,400 participants. We will demonstrate how we utilise Hierarchical Bayesian Logistic Regression to separate users’ aesthetic biases from their genuine semantic understanding. We will reveal concrete, data-driven findings on what truly matters in emoji design. For example, a candy emoji fundamentally requires stripes; extra details like white gloves can unexpectedly harm understanding; and the precise shape and placement of a sweat drop dictates emotional intent. Attendees will see examples of vendor designs we have influenced and new emoji codepoints we have added as a direct result of this applied research.
How this connects to ‘Unicode in the World’ theme
Global public infrastructure
Emoji are universally relied-upon for conveying tone, emotion, and body language in digital text. Just as Unicode ensures a character encodes correctly across devices, the ESR’s interoperability research ensures the emotional intent encodes correctly. Because cross-platform visual variation can lead to real-world miscommunication, this research is the mechanism by which we safeguard emoji as a reliable, standardised public utility.
Opening doors to new audiences
This research intersects linguistics, digital humanities research, and UI/UX design. By showcasing our experimental methodologies and empirical findings, this session intentionally invites linguists and researchers to see how human-centered data is evaluated at scale, while providing designers and typographers with actionable insights on how their cross-vendor aesthetic choices impact human connection.
What we expect audience members to gain from it
Designers & Typographers: A data-backed understanding of which visual elements actually drive user comprehension. Attendees will learn to distinguish between critical semantic markers (e.g. structural details, directionality, and microexpressions) and flexible stylistic flair (e.g. color palettes and shading). Linguists & Researchers: An inside look at the methodology behind Unicode’s decision-making. We will explain how we design large-scale linguistic experiments, use Hierarchical Bayesian Logistic Regression to adjust for vendor style bias, and mathematically account for CLDR label ambiguity. The broader Unicode Community: A behind-the-scenes look at how standardisation continues long after a new codepoint is approved. Attendees will see real-world examples of how empirical research has directly led major vendors to update their designs to better serve the communicative needs of the public without dictating distinctive artistic style.