Overview of SpeechMATICS at IberLEF 2026: Speech Multimodal Audio-Text Irony Classification in Spanish

Resumen

This paper presents an overview of the SpeechMATICS 2026 shared task, organized as part of IberLEF 2026. The main goal of the task is to advance research on automatic irony detection in spontaneous spoken Spanish from both textual and multimodal perspectives. The task is divided into two independent subtasks. The first subtask focuses on irony detection using only manual transcriptions of audio fragments. The second subtask introduces a multimodal setting in which systems can combine textual and acoustic information. SpeechMATICS is based on MESironic, a multimodal corpus of short spoken fragments with manually verified transcriptions and irony annotations. A total of 14 teams registered for the task, 4 submitted official runs, and 4 submitted working notes describing their systems. This overview describes the task, dataset, evaluation protocol, participant approaches, official results, and main findings from this first edition.