Mapping Political Sentiment Trajectories in Pakistan’s 2023 Election Manifestos: A Corpus-Based Comparative Analysis of Urdu and English Media Discourse

Abstract

This research investigates the sentiment trajectory embedded in Pakistan's 2023 election manifestos, analyzing both Urdu and English media representations. The focus lies in understanding how political narratives are framed and perceived differently in these two linguistic domains. Given the increasingly polarized political landscape in Pakistan, it is crucial to examine how political discourse is shaped and communicated in both native and international languages. Previous research on political manifestos has highlighted the growing role of media in shaping electoral outcomes (Tufekci, 2014; Thorson & Wells, 2016). Our study employs a corpus- based sentiment analysis approach, utilizing machine learning models (e.g., BERT, Naïve Bayes) to extract sentiment trajectories across election manifestos, which are analyzed in the context of both Urdu and English media. The study finds significant differences in sentiment, with Urdu manifestos predominantly reflecting nationalistic and populist tones, while English manifestos focus more on economic and global concerns. The findings suggest that language plays a pivotal role in shaping political sentiment and that media outlets' linguistic choices influence public reception and electoral success. This study contributes to understanding political communication in a multilingual society and offers insights for both political analysts and media practitioners.

Keywords: Corpus-based analysis, Sentiment trajectory, Election manifestos, Urdu media, English media, Pakistan politics, Political discourse