NLP and Computational Stylistics for Iraqi Literature and Media: Evidence Map (2023–2026; includes one diachronic study covering 1980–2025), Methodological Standards, and a Responsible Roadmap

Abstract

This review synthesizes how Natural Language Processing (NLP) and computational stylistics are being used to analyze Iraqi literature, media, and culturally salient text streams. Because Iraqi Arabic is low-resource and highly variable, the paper emphasizes evaluation rigor, dataset governance, and interpretability for humanities-facing claims. We combine (i) a structured task taxonomy (sentiment/affect, topic discovery, stylometry, censorship detection, diachronic semantic change, dialect translation, and ancient text processing) with (ii) a title-coded evidence map of a curated seed corpus (N=35) drawn from Iraqi Literary and Cultural Review (ILCR) and Alnoor Journal for Humanities (JNH). The synthesis highlights recurring technical risks (OCR noise, code-switching, domain shift, small-sample instability) and proposes a practical roadmap: shared benchmarks for Iraqi dialect and genre coverage, transparent annotation protocols, robust metrics with uncertainty reporting, and privacy-preserving release strategies for sensitive political data.