Pre-Islamic Poetry between Classical Commentary and Machine Analysis: A Linguistic Study Using Deep Learning

Abstract

This study aims to present a linguistic comparison between human and machine analysis of pre-Islamic poetic texts, by employing deep learning techniques in Arabic language processing. Advanced language models were used to analyze selected verses from the poetry of Imru’ al-Qays, in order to assess their ability to handle the linguistic and artistic features of this literary genre.

The machine-based analysis involves computational tools that process morphological, syntactic, and semantic levels, classify poetic themes, and attempt to identify rhetorical phenomena. In contrast, human analysis relies on classical commentaries that offer interpretive readings and reveal the symbolic and cultural dimensions of the texts, drawing on linguistic and contextual knowledge that still surpasses the capabilities of current AI models.

The findings indicate that deep learning models are effective in handling the formal linguistic structure of the texts and identifying general poetic themes. However, they struggle to grasp complex metaphors, cultural references, and contextual meanings that require deeper background knowledge.

The study concludes that integrating human and machine analysis enriches the understanding of heritage texts and highlights the need to develop more specialized language models for processing classical Arabic literature. It also emphasizes the importance of expanding training datasets to include annotated heritage texts, enabling artificial intelligence to achieve greater accuracy and depth in analyzing culturally and historically rich material.