Applications of NLP in Computational Poetics and Literary Analysis

The convergence of Natural Language Processing (NLP) and computational creativity has catalyzed a transformative shift in how poetic and literary forms are generated, interpreted, and evaluated through artificial intelligence. This systematic review critically examines the applications of NLP in computational poetics and literary analysis, surveying research outputs from 2000 to 2025 across peer-reviewed journals, conference proceedings, and technical reports. Through a methodical synthesis of 115 scholarly works, this review identifies core advances in the computational modeling of poetic structure, figurative language, literary style, and algorithmic interpretation of texts. The study categorizes developments into five principal domains: (1) automated poetry generation using rule-based, probabilistic, and deep learning models; (2) structural and metrical analysis of poetic forms, including rhyme, rhythm, and lineation through syntactic parsing; (3) computational interpretation of metaphor, symbolism, and affect using sentiment analysis and semantic networks; (4) authorial style emulation and genre classification through stylometry and neural embeddings; and (5) large-scale literary analysis through topic modeling, narrative extraction, and discourse segmentation. Recent advances in transformer-based models such as GPT-4, T5, and BERT have enabled significant gains in linguistic fluency and stylistic imitation in generated texts. However, the review identifies persistent limitations regarding semantic originality, cultural depth, and long-form narrative coherence—aspects crucial to authentic literary creativity. Moreover, while AI-generated poetry can mirror formal constraints and emotional cues, it frequently lacks the intentionality, irony, and conceptual depth of human-authored verse. The review also explores the emerging field of human-AI literary co-creation, where language models function as collaborators or assistants in poetic composition, and NLP tools support interpretive and pedagogical engagement with literature. Ethical considerations around algorithmic authorship, intellectual property, dataset bias, and cultural homogenization are discussed in light of their implications for literary diversity and critical discourse. The review emphasizes the need for cross-disciplinary evaluation frameworks that align computational creativity with humanistic criteria such as originality, metaphorical insight, and symbolic resonance. Ultimately, this study positions NLP not merely as a tool for automation, but as a medium of augmentation—reshaping the boundaries of literary production and analysis. It concludes by outlining an interdisciplinary research agenda that integrates linguistic computing, literary theory, cultural studies, and ethical AI to foster richer and more inclusive forms of digital creativity.