Call for Participation for MultiPRIDE task at EVALITA 2026.
With this task, we invite researchers, practitioners and students to explore features and issues related to reclaimed language in the LGBTQ+ community.
In particular, we suggest participants to focus on both the textual content of the inputs (such as arguments, slurs, denigratory words, self-labeling, figures of speech), and the contextual information that can be inferred from the users’ profile (when available),
such as their being part of the LGBTQ+ community.
This is a binary classification task, in which systems must classify whether a term related to LGBTQ+ context in a sentence is used with a reclamatory intent or not.
Can reclaimed slurs be correctly detected by LLMs?
We propose two different tasks that can be addressed according to a constrained or unconstrained approach:
• Task A - Textual Content: participants are provided only with the textual content of the tweet.
• Task B - Contextual Content: in addition to the textual content of the tweet, participants can use contextual information related to the author’s profile, such as their biography (when available).
MultiPRIDE is a multilingual task and involves one or more languages among Italian, Spanish and English data. Although not mandatory, participants are encouraged to foster cross-linguistic analysis.
Important Dates:
• September 22, 2025: release of training data
• November 21, 2025: Call for Interest deadline
• November 27 – December 4, 2025: evaluation window
• December 15, 2025: results to participants
• January 9, 2026: submission of participants’ report
• February 7, 2026: reviews to participants
• February 16, 2026: camera ready
• February 26 – 27, 2026: EVALITA workshop (Bari)
The MultiPRIDE team
Lia Draetta, Chiara Ferrando, Marco Madeddu, Mae Sosto, Viviana Patti, Paolo Rosso, Cristina Bosco, Jacinto, Estrella Gualda