Framing the Fray: Conflict Framing in Indian Election News Coverage

Tejasvi Chebrolu, Rohan Modepalle, N. Harsha Vardhan, Ponnurangam Kumaraguru, Ashwin Rajadesingan

TL;DR

This paper investigates how Indian news outlets frame election coverage, specifically focusing on "conflict frames" that emphasize adversarial interactions. They found that TV media, rather than ideological bias, drives the use of these frames, and that while attacks are widely reported, substantive policy issues are often underrepresented.

What Problem Does This Paper Address?

News framing plays a crucial role in shaping public perception of elections. While conflict frames can increase engagement, they may also distract from important policy discussions and contribute to political polarization. This paper addresses the question of how prevalent conflict frames are in Indian election news, whether this prevalence is linked to media bias or modality, and how it compares to the reporting of substantive issues. Specifically, it examines whether news outlets overemphasize conflict over policy discussions, a common concern in political journalism.

Methods

The researchers employed a multi-stage approach:

  1. Conflict Frame Detection: They fine-tuned a pre-trained DistilBERT model, a transformer-based neural network, to classify news headlines as either containing a conflict frame or not. This model was trained on a custom dataset of 860 annotated Indian election news headlines.
  2. Aggressor/Target Identification: For headlines identified as conflict frames, they used GPT-4 Turbo to extract the "aggressor" (the entity initiating the conflict) and the "target" (the entity being attacked).
  3. Issue Extraction and Comparison: They collected campaign speeches and matched them with news articles published within three days. Issues mentioned in speeches were manually extracted and compared to the issues covered in the corresponding news articles to assess under- or over-reporting.

Datasets

  • News Articles: A dataset of 69,400 English-language news headlines from seven major Indian news outlets (The Hindu, The Times of India, The Indian Express, NDTV, India Today, Republic World, and Zee News) covering the 2014 and 2019 Indian general elections.
  • Conflict Frame Training Data: A custom dataset of 860 annotated news headlines used to train the conflict frame classifier.
  • Campaign Speeches: A dataset of 224 campaign speeches from BJP and INC leaders, some of which were translated from Hindi using GPT-4 Turbo.

Key Insights and Ideas

  • TV Trumps Bias: The study found that media modality (TV vs. print) is a significant predictor of conflict frame usage, with TV outlets being twice as likely to use them. This suggests that the visual and fast-paced nature of television news might favor conflict-driven narratives.
  • National Parties Dominate Conflict: Headlines mentioning national parties are significantly more likely to employ conflict frames compared to those mentioning regional parties. This could amplify the perception of national politics as inherently conflictual.
  • Attacks are News, Policy is Not: While news outlets consistently report on attacks made by politicians (especially when national parties are involved), they significantly under-report on substantive electoral issues like farmers' welfare or infrastructure, even when these issues are prominent in campaign speeches.
  • Ideological Bias Less Influential: Surprisingly, the researchers found little evidence that the ideological bias of news outlets significantly correlates with their use of conflict frames. Even a pro-BJP outlet (Republic World) reported a substantial proportion of headlines where the NDA was being attacked.

Key Results

  • Conflict Frame Prevalence: Approximately 19.29% of the analyzed news headlines were classified as containing a conflict frame.
  • Media Modality Impact: TV-based news outlets were found to be twice as likely to use conflict frames compared to print-based outlets.
  • Party Type Impact: Headlines mentioning national parties were significantly more likely to use conflict frames than those mentioning regional parties.
  • Issue Reporting Discrepancy: When reporting on campaign speeches, news articles mentioned the opposition party in 88% of cases, but only mentioned substantive issues like farmers' issues in about 30% of cases.

Limitations and Future Work

The authors acknowledge several limitations:

  • Focus on Conflict Frames: The analysis was limited to conflict frames, and future work could incorporate other frames like strategy or game frames.
  • Headline-Centric Analysis: While headlines are influential, the body of articles and multimedia elements also contribute to framing and warrant further investigation.
  • Data Source Limitations: The campaign speech dataset was limited to national parties, and the news article dataset excluded regional language publications.
  • Contextual Specificity: The research is contextually focused on India, and comparative analyses with other democratic systems could offer broader insights.

Future work could also explore the impact of media ownership, market share, and the integration of multimedia content. Additionally, a comparative analysis of headline framing across media outlets with varying political orientations could provide more nuanced understanding of media bias.

Summary updated

BibTeX
@inproceedings{chebrolu2025framing,
  author = {Tejasvi Chebrolu and Rohan Modepalle and N. Harsha Vardhan and Ponnurangam Kumaraguru and Ashwin Rajadesingan},
  title = {{Framing the Fray: Conflict Framing in Indian Election News Coverage}},
  booktitle = {WebSci},
  year = {2025},
  doi = {10.1145/3717867.3717900},
}
Figure