Abstract:
This exploratory paper examines whether claims of emoji universality hold for AI-generated emoji renditions of a single Hausa folktale produced by three Artificial Intelligence (AI) tools. We have observed that published emoji research on African languages appears limited, and Hausa, one of the largest African languages with an estimated 94 million first- and second-language speakers (Eberhard et al., 2025), is no exception. Using a convergent mixed-methods design that combines computational description of the emoji sequences with evaluation by 20 native Hausa speakers, the study compares how three AI tools (ChatGPT, Perplexity AI, and DeepSeek) render the folktale as emoji-based narratives. The findings reveal substantial variation in emoji representation across AI tools, highlighting challenges in cross-cultural and cross-linguistic emoji interpretation. The renditions varied markedly in length, emoji diversity (25 to 44 distinct emojis), character encoding, and narrative segmentation. Participants rated DeepSeek highest for interpretability (4.2 out of 5) and cultural appropriateness (4.0), rated ChatGPT highest for emotional accuracy (4.1), and preferred DeepSeek overall (45%, compared with 35% for ChatGPT and 20% for Perplexity AI). Because these findings are limited to three systems and one folktale, they caution against assuming emoji renditions are universally interpretable and instead argue that they are not. The study offers preliminary evidence for research on digital communication in African contexts and argues for emoji research attentive to the linguistic and cultural specificities of African languages. It concludes with specific recommendations for future research, including Hausa-specific datasets, annotation guidelines, and human evaluation protocols.
DOI: https://www.doi.org/10.57233/djfafais.2026.v22i02.010
author/Abdullahi Aliyu, Aisha Umar Adam and Aishatu Shehu Maimota
journal/Degel 22(2) | October 2026 |

