Search papers, labs, and topics across Lattice.
This study investigates the impact of fine-tuning on large language models (LLMs) using culturally rich Arabic datasets to enhance figurative language understanding. The results show that fine-tuning on poetry significantly improves idiom comprehension, while cultural fine-tuning negatively affects proverb interpretation accuracy, indicating a complex relationship between cultural knowledge and figurative language processing. Additionally, the research highlights that while multilingual models exhibit greater adaptability, fine-tuning can destabilize historically grounded factual knowledge, complicating the transfer of cultural insights across figurative domains.
Fine-tuning on poetry boosts idiom comprehension in LLMs, but cultural fine-tuning paradoxically undermines proverb interpretation accuracy.
Figurative language is deeply culturally embedded; fluent use requires not just linguistic competence but cultural immersion. We ask whether LLMs can learn this link: does fine-tuning on cultural data improve figurative language understanding, and vice versa? We conduct a systematic study across four models (ALLaM-7B, Fanar-1-9B, Qwen3-8B, Llama-3.1-8B) and six Arabic datasets spanning cultural commonsense, proverbs, and poetry across diverse dialects and regions. Fine-tuning on poetry improves idiom comprehension (+2.33%, p<0.05), a gain our ArabicMMLU control does not reproduce, indicating that it stems from figurative content rather than Arabic language adaptation and pointing to a sensitivity to non-literal meaning that transfers across figurative types. Cultural fine-tuning, by contrast, lowers proverb-interpretation accuracy in both Arabic-centric models. Transfer between the two domains is otherwise indistinguishable from noise, with Arabic models frequently regressing after fine-tuning, suggesting prior saturation of relevant knowledge, while multilingual models show greater adaptation headroom. Error analysis further reveals that fine-tuning reinforces experiential cultural knowledge while destabilizing historically grounded factual knowledge. Our findings suggest that the relationship between culture and figurative language, though conceptually natural, is not straightforwardly captured through fine-tuning alone.
