Paper accepted at EACL 2026: "Controlling Reading Ease with Gaze-Guided Text Generation"

• We used a model that predicts eye movements while reading to steer an off-the-shelf language model.
• This approach allows us to generate texts that elicit specific reading patterns (e.g., faster or slower reading).
• Our human evaluation showed that the resulting texts are also perceived as being easier or harder to read.

We will present our submission at the 19th Conference of the European Chapter of the Association for Computational Linguistics in Rabat (Morocco) on March 25-27, 2026.

Read the paper here and check out the eye-tracking dataset here.