Yor-Sarc: A gold-standard dataset for sarcasm detection in a low-resource African language

Toheeb A. Jimoh, Tabea De Wille, Nikola S. Nikolov · 2026

Read Yor-Sarc: A gold-standard dataset for sarcasm detection in a low-resource African language by Toheeb A. Jimoh, Tabea De Wille, Nikola S. Nikolov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-18964.txt.

Description

Sarcasm turns on the gap between literal and intended meaning, and that gap is culture-bound. For Yorùbá, a tonal Niger-Congo language with over fifty million speakers, no gold-standard dataset existed. Toheeb Aduramomi Jimoh, with two co-authors, presents Yor-Sarc in 2026, the first such dataset for sarcasm detection in Yorùbá.

The dataset holds hundreds of instances marked by three native speakers of different dialects. Its protocol is designed specifically for Yorùbá sarcasm, taking culture into account, with context-sensitive interpretation and community-informed guidelines. Agreement ranges from substantial to almost perfect, with a large majority of unanimous consensus. Remaining majority-agreement cases are kept as soft labels for uncertainty-aware modelling. Yor-Sarc should aid research on meaning interpretation and culture-aware NLP for low-resource African languages.

Catalog record

CREATORToheeb A. Jimoh, Tabea De Wille, Nikola S. Nikolov
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE31 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-18964

Browse the full library · About the Tower of Babel