Meet Paula: The AI Avatar Learning to Sign With Emotion

Most people picture sign language avatars as robotic, stiff, and a little lifeless. That’s precisely the problem a European research project called EASIER set out to solve, one facial expression and one well-timed pause at a time.

A Different Approach to Avatar Technology

EASIER was a three-year Horizon 2020 project, funded with a €4 million EU grant, running from January 2021 to December 2023. Led by a 14-partner European consortium with scientific direction from Dr. Eleni Efthimiou of the Athena Research Center, the project focused on seven sign languages: German, Swiss German, French, British, Greek, Italian, and Dutch Sign Language, paired with their corresponding spoken languages.

At the center of the project sits Paula, EASIER’s signing avatar. What makes Paula different from earlier generations of sign avatars is the project’s explicit focus on affective expression and coherent prosody, in plain terms, making sure the avatar’s signing carries emotional tone and natural rhythm, not just technically correct handshakes.

Built for Everyday Situations

EASIER wasn’t designed as a lab curiosity. The project envisioned practical, everyday use cases: translating written text on a webpage into sign language on the fly, helping schedule a quick low-stakes appointment, or supporting sign language professionals like teachers and interpreters in their daily work. The goal throughout was creating a productivity tool that fits into real workplace and everyday-life situations, not just a demo.

A Deaf-Involved Design Process

Like its sister project SignON, EASIER built in collaboration with Deaf community members throughout the design and validation process, rather than testing with Deaf users only after the technology was finished. That approach reflects a growing consensus in sign language technology research: avatars and translation tools work better, and earn more trust, when Deaf people help shape them from the start.

Why Emotional Nuance Matters So Much Here

For hearing viewers, this might sound like a nice-to-have feature. For sign language users, it’s closer to essential. Facial expression, timing, and rhythm aren’t decoration in signed languages; they carry grammatical and emotional information the same way tone of voice does in spoken language. An avatar that gets the handshakes right but misses the expression is, in a real sense, still only telling half the story.

The Bigger Picture

EASIER is part of a broader wave of EU-funded sign language technology research, alongside projects like SignON, all wrestling with the same core challenge: building translation tools that don’t just move hands correctly, but actually communicate the way real signers do.

Interested in International Sign or the global Deaf community? Follow OneDeafWorld.com for more coverage of IS, accessible technology, and events from around the world.