Computational Identification of Gender-Based Violence Discourse in Pakistani English Newspapers Using Corpus Linguistics and NLP-Based Semantic Analysis

Abstract

Although computational approaches are common in media research, their application to the study of gender-based violence remains limited, particularly when combined with critical discourse perspectives. This study presents a hybrid analytical framework that integrates corpus linguistics with Feminist Critical Discourse Analysis (FCDA) to examine how gender-based violence is represented in Pakistani English-language newspapers. The corpus comprises 500 news reports published between 2017 and 2025, containing approximately 1.37 million words collected from five leading national newspapers. W-Matrix was employed to conduct keyword extraction, semantic annotation, and concordance analysis, providing an objective basis for identifying recurring lexical and semantic patterns. These computational findings were subsequently interpreted through FCDA to explore the ideological meanings embedded in media discourse (Khan, 2025). The analysis reveals persistent linguistic patterns associated with honour killings, rape, domestic violence, acid attacks, dowry-related abuse, and property disputes. Across the corpus, women are predominantly portrayed as victims, whereas perpetrators are frequently backgrounded through passive constructions and event-oriented reporting. The findings demonstrate that combining computational corpus analysis with critical discourse interpretation offers a robust approach to examining gendered representations in large news datasets. The proposed framework contributes to computational media analytics by providing a systematic method for investigating socially significant discourse while preserving the contextual depth required for critical interpretation.