Elamur Research⌕
☰
home / papers / 2026 / How English Print Media Frames Human-Elephant Conflicts in India

How English Print Media Frames Human-Elephant Conflicts in India

Paper · 2026 · Bonala Sai Punith, Salveru Jayati, Garima Shakya et al.

Summary

The paper asks how Indian English-language newspapers depict elephants in reports on human–elephant conflict (HEC). It also asks whether that depiction leans negative at scale. Ecological causes of HEC are well studied, but media portrayal had barely been examined computationally. The authors used MediaCloud URLs and then scraped, filtered and cleaned the results. This produced a corpus of 1,968 full articles (28,986 sentences) from one widely read national outlet, covering January 2022 to September 2025. Each article is scored as a whole document by several kinds of system: long-context transformers, instruction-tuned LLMs, rule-based tools and a new domain lexicon, the Negative Elephant Portrayal Lexicon (NEPL). LLMs also select the sentences that justify each sentiment decision. No large gold standard exists, so evaluation relies on agreement between models and on comparison with expert annotators. Negative portrayals clearly dominate, and words of aggression and fear appear in most articles. Government advisories against sensational wording seem to have changed little. This matters because such framing can harden public hostility and weaken coexistence-oriented conservation policy.

Key points

  • Corpus: 1,968 HEC articles, 28,986 sentences, Jan 2022–Sep 2025, from one major English outlet. URLs came from MediaCloud and were followed by additional scraping and cleaning.
  • The pipeline combines long-context transformers, LLMs, rule-based systems and the expert-reviewed NEPL. NEPL covers literal, metaphorical and fear-inducing expressions.
  • Bias is analysed across the timeline and split by fatal vs. non-fatal incidents. 101 articles were compared against expert annotations, and inter-expert agreement was only moderate.
  • Rationale sentences from the LLMs were scored against expert-chosen sentences with BLEU and ROUGE-L. Gemini overlapped more with Expert 2 than with Expert 1, and agreement on reasoning was moderate overall.
  • Advisories issued by MoEF in 2021 and by the Chhattisgarh Forest Department in 2025 show little visible effect on reporting style.

Interesting ideas

  • I1. Negativity is often implicit: the tone stays neutral, but verb choice gives the animal intent, which turns an ecological event into an attack-and-defence story.
  • I2. When no gold labels exist, the paper uses agreement between diverse models plus expert review of their justifications as an indirect quality signal. It also states openly that this signal is weaker than accuracy or F1.
  • I3. Models and experts often agreed on the drivers of a sentiment judgement yet picked different sentences as evidence. This shows that lexical overlap (BLEU/ROUGE-L) undervalues rationales that are related in meaning but not in wording.
  • I4. The domain lexicon was built from three categories (plain lexical, metaphorical, fear-inducing) and then vetted by experts. This gives a transparent, auditable cue list instead of an opaque classifier.
  • I5. Splitting the analysis by fatal vs. non-fatal incidents tests whether negative wording simply follows the severity of the event.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): Check which agent a verb like "attack" or "rampage" makes the deliberate actor, and flag where intentional-agent roles go to non-human participants while causal context is missing.
  • I2 (learn): This mirrors silver data from several witness parsers. Adding a human check of the justification behind agreed outputs would catch cases where the witnesses agree but are all wrong.
  • I3 (answer): When scoring the evidence behind QA answers, compare the frames or graph nodes the evidence supports rather than exact sentence overlap.
  • I4 (world): Store evaluative and connotation tags on lexicon or ontology entries as explicit compiled knowledge. Each portrayal judgement can then be traced to a named entry.
  • I5 (learn): Use event severity as a control condition, so that tone detectors which only react to casualty mentions can be told apart from those that detect framing.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of DubaiDubai · United Arab Emirates2 works · 4 citations

Also on the byline: Indian Institute of Technology Palakkad.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksSocial Sciences · 1,772 worksCommunication · 97 worksEnvironmental Science · 9 worksEcology · 2 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Bonala Sai Punith, Salveru Jayati, Garima Shakya and Shubham Kumar Nigam. 2026. How English Print Media Frames Human-Elephant Conflicts in India. arXiv:2604.21496.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2604.21496 · arXiv 2604.21496

@misc{punith2026how,
  title = {{How English Print Media Frames Human-Elephant Conflicts in India}},
  author = {Bonala Sai Punith, Salveru Jayati, Garima Shakya and Shubham Kumar Nigam},
  year = {2026},
  doi = {10.48550/arxiv.2604.21496},
  eprint = {2604.21496},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2604.21496}
}

References

30 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bai, Y.; Zhang, Z.; Cui, C.; and Team, Q. 2024. Qwen2: A Family of Powerful Transformer Models. arXiv preprint arXiv:2404.02076. ↗
  2. Bal, P.; Nath, C. D.; Nanaya, K. M.; Kushalappa, C. G.; and Garcia, C. 2011. Elephants also like coffee: trends and drivers of human–elephant conflicts in coffee agroforestry landscapes of Kodagu, Western Ghats, India. Environmental management, 47(5): 789–801. ↗
  3. Barbosa, J. S.; Gondhali, U.; Petrossian, G.; Sharma, K.; Chakraborty, S.; Jacquet, J.; and Freire, J. 2025. A Cost-Effective LLM-based Approach to Identify Wildlife Trafficking in Online Marketplaces. Proc. ACM Manag. Data, 3(3). ↗
  4. Barua, M. 2010. Whose issue? Representations of human-elephant conflict in Indian and international media. Science Communication, 32(1): 55–75. find
  5. Beltagy, I.; Peters, M. E.; and Cohan, A. 2020. Longformer: The Long-Document Transformer. arXiv preprint arXiv:2004.05150. ↗
  6. Indian Readership Survey 2019 Q4. Last accessed 28 November 2025. find
  7. de Nazareth, M.; and Nagarathinam, S. 2012. Human elephant conflict and the role of print media. Gajah, 38. find
  8. DeepMind, G. 2023. Gemini: A Family of Highly Capable Multimodal Models. https://deepmind.google/technologies/gemini/. Accessed: 2025-01-20. ↗
  9. Down To Earth, . 2025. When headlines hurt: How media fuels the human-elephant conflict. https://www.downtoeart h.org.in/wildlife-biodiversity/when-headlines-hurt-how-media-fuels-the-human-elephant-conflict. Accessed: 2026-01-15. ↗
  10. Ekuban, C. A. A. 2025. Time Series Analysis of Human-Elephant Conflict Mortality Rates: A Case Study of the Coimbatore Forest Area. ↗
  11. Goswami, V. R.; Medhi, K.; Nichols, J. D.; and Oli, M. K. 2015. Mechanistic understanding of human–wildlife conflict through a novel application of dynamic occupancy models. Conservation Biology, 29(4): 1100–1110. ↗
  12. Hutto, C.; and Gilbert, E. 2014. VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. In Proceedings of the International AAAI Conference on Weblogs and Social Media. find
  13. Jurafsky, D.; and Martin, J. H. 2000. Speech and Language Processing. Prentice Hall. find
  14. Kuswanda, W.; Garsetiasih, R.; Gunawan, H.; Situmorang, R. O. P.; Hutapea, F. J.; Kwatrina, R. T.; Karlina, E.; Atmoko, T.; Zahrah, M.; Takandjandji, M.; et al. 2022. Can humans and elephants coexist? A review of the conflict on Sumatra Island, Indonesia. Diversity, 14(6): 420. find
  15. LaDue, C. A.; Vandercone, R. P.; and Snyder, R. J. 2025. Representations of human–elephant interactions on YouTube: Analyzing content and engagement to inform sustainable practices for wildlife tourism. Biological Conservation, 308: 111240. Lok-Sabha, U. Q. N. . 2024. GOVERNMENT OF INDIA MINISTRY OF ENVIRONMENT, FOREST AND CLI-MATE CHANGE. Last accessed 28 November 2025. ↗
  16. Mammola, S.; Malumbres-Olarte, J.; Arabesky, V.; Barrales-Alcalá, D. A.; Barrion-Dupo, A. L.; Benamú, M. A.; Bird, T. L.; Bogomolova, M.; Cardoso, P.; Chatzaki, M.; et al. 2022. The global spread of misinformation on spiders. Current Biology, 32(16): R871–R873. ↗
  17. Maurer, G.; Chandelier, M.; Mulot, B.; and Gimenez, O. 2024a. Polarized media coverage of conflicting, yet emblematic species: The ambivalent portrayal of the Asian elephant. Biological Conservation, 289: 110391. ↗
  18. Maurer, G.; Chandelier, M.; Mulot, B.; and Gimenez, O. 2024b. Polarized media coverage of conflicting, yet emblematic species: The ambivalent portrayal of the Asian elephant. Biological Conservation, 289: 110391. ↗
  19. McCombs, M. 2005. A look at agenda-setting: Past, present and future. Journalism studies, 6(4): 543–557. ↗
  20. Menon, V.; and Tiwari, S. K. 2019. Population status of Asian elephants Elephas maximus and key threats. International Zoo Yearbook, 53(1): 17–30. ↗
  21. Mumby, H. S.; and Plotnik, J. M. 2018. Taking the elephants’ perspective: remembering elephant behavior, cognition and ecology in human-elephant conflict mitigation. Frontiers in Ecology and Evolution, 6: 122. ↗
  22. Muter, B. A.; Gore, M. L.; Gledhill, K. S.; Lamont, C.; and Huveneers, C. 2013. Australian and US news media portrayal of sharks and their conservation. Conservation Biology, 27(1): 187–196. find
  23. Nanni, V.; Caprio, E.; Bombieri, G.; Schiaparelli, S.; Chiorri, C.; Mammola, S.; Pedrini, P.; and Penteriani, V. 2020. Social media and large carnivores: Sharing biased news on attacks on humans. Frontiers in Ecology and Evolution, 8: 71. ↗
  24. Okpala, I.; Rodriguez, G. R.; Han, C.; Meierhofer, M.; Mammola, S.; Halse, S.; Kropczynski, J.; and Johnson, J. 2024. A framework for perception analysis of social media data during disease outbreaks: uncovering patterns of resentment towards bats. In 57th Hawaii International Conference on System Sciences, 2475–2484. Hawaii International Conference on System Sciences. find
  25. Prakash, S. L.; Perera, P.; Newsome, D.; Kusuminda, T.; and Walker, O. 2019. Reasons for visitor dissatisfaction with wildlife tourism experiences at highly visited national parks in Sri Lanka. Journal of Outdoor Recreation and Tourism, 25: 102–112. ↗
  26. Rathi, P. 2025. Navigating the Landscape: Challenges and Opportunities of AI in Wildlife Conservation. AI and Machine Learning Techniques for Wildlife Conservation, 159– 182. find
  27. Roy, K.; Pandey, R. K.; Athira, N.; Dutta, A.; Mittal, D.; Nigam, P.; Nath, A.; and Habib, B. 2025. Long-term trends in human fatalities from human–elephant conflict in Chhattisgarh, India. Scientific Reports, 15(1): 27360. ↗
  28. Santangeli, A.; Mammola, S.; Nanni, V.; and Lambertucci, S. A. 2024. Large language models debunk fake and sensational wildlife news. Integrative conservation, 3(2): 127– 133. ↗
  29. Shanahan, E. A.; McBeth, M. K.; and Hathaway, P. L. 2011. Narrative policy framework: The influence of media policy narratives on public opinion. Politics & Policy, 39(3): 373– 400. ↗
  30. Siebert, O. 2020. Sentiment Classification using RoBERTa. HuggingFace Model Card. https://huggingface.co/siebert/s entiment-roberta-large-english. The New Indian Express, . 2021. MoEF to all state PCCF: tell media ’not’ to portray elephants negatively with harsh remarks. Last accessed 28 November 2025. The New Indian Express, . 2025. Not ‘terrorists’ or ‘murderers’: Avoid using harsh words for elephants, Chhattisgarh forest dept tells media. Last accessed 28 November 2025. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/punith-2026-english-print-media-frames-human/