Computationally Distinguishing Quran and Pre-Islamic Arabic Poetry

The literary uniqueness and coherence of the Quran have been debated for centuries, with some Western scholars claiming that it draws from pre-Islamic Arabic poetry and lacks coherence, while others refuting these claims. However, this entire body of work relies mainly on qualitative analyses of selected verses. Here, we employed modern NLP techniques to systematically compare and contrast all verses of the Quran and pre-Islamic poetry. Specifically, we applied two BERT-based pre-trained models on Quran and pre-Islamic poetry verses to get embeddings, preprocessed them, and applied four clustering algorithms. Our results show that Quran (even Makki verses of the Quran - which are more poetic) and poem verses tend to form distinct clusters. Thus, our quantitative approach addresses gaps in the literature, offering fresh insights into long-standing debates. To our knowledge, this work is the first of its kind.