arxiv: 2604.21496 · v2 · submitted 2026-04-23 · 💻 cs.AI · cs.CL· cs.CY

Recognition: no theorem link

How English Print Media Frames Human-Elephant Conflicts in India

Bonala Sai Punith, Garima Shakya, Salveru Jayati, Shubham Kumar Nigam

Authors on Pith no claims yet

Pith reviewed 2026-05-14 22:02 UTC · model grok-4.3

classification 💻 cs.AI cs.CLcs.CY

keywords human-elephant conflictmedia framingsentiment analysiswildlife conservationnatural language processingIndiacomputational social science

0 comments

The pith

English print media in India frames human-elephant conflicts using predominantly fear-inducing and aggression-related language.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper conducts the first large-scale computational study of how English news outlets in India portray human-elephant conflicts. By analyzing nearly 2000 full-length articles with a combination of transformers, large language models, and a custom lexicon, it finds that negative portrayals of elephants dominate the coverage. Such framing can shape public attitudes and potentially undermine conservation efforts aimed at coexistence. The study provides a scalable method to assess media narratives on wildlife issues and releases its resources for further use.

Core claim

The analysis of 1,968 news articles reveals a dominance of fear-inducing and aggression-related language in reporting on human-elephant conflicts, quantified through a multi-model framework that extracts sentiment and linguistic patterns contributing to negative portrayals.

What carries the argument

A multi-model sentiment framework that integrates long-context transformers, large language models, and a domain-specific Negative Elephant Portrayal Lexicon to quantify negative portrayals and extract rationale sentences.

Load-bearing premise

The multi-model framework combining transformers, LLMs, and the custom lexicon accurately captures negative portrayals without introducing bias from the models or lexicon.

What would settle it

A large-scale manual annotation of a sample of articles by domain experts showing significantly lower negative sentiment than the automated scores.

Figures

Figures reproduced from arXiv: 2604.21496 by Bonala Sai Punith, Garima Shakya, Salveru Jayati, Shubham Kumar Nigam.

**Figure 2.** Figure 2: A news article stating the advisory by the Ministry [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Figure 3: A news article stating the advisory by Chhattisgarh [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗

**Figure 4.** Figure 4: Presence of NEPL categories across the articles. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗

**Figure 6.** Figure 6: Comparison of proportion of labels across models. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗

**Figure 7.** Figure 7: Comparison of negativity rates across LLM, trans [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗

**Figure 9.** Figure 9: Vader Compound Scores vs number of articles [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗

**Figure 10.** Figure 10: Output of Regex on articles having Vader Com [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗

**Figure 11.** Figure 11: Word cloud to show sentiment distribution [PITH_FULL_IMAGE:figures/full_fig_p013_11.png] view at source ↗

read the original abstract

Human-elephant conflict (HEC) is rising across India as habitat loss and expanding human settlements force elephants into closer contact with people. While the ecological drivers of conflict are well-studied, how the news media portrays them remains largely unexplored. This work presents the first large-scale computational analysis of media framing of HEC in India, examining 1,968 full-length news articles consisting of 28,986 sentences, from a major English-language outlet published between January 2022 and September 2025. Using a multi-model sentiment framework that combines long-context transformers, large language models, and a domain-specific Negative Elephant Portrayal Lexicon, we quantify sentiment, extract rationale sentences, and identify linguistic patterns that contribute to negative portrayals of elephants. Our findings reveal a dominance of fear-inducing and aggression-related language. Since the media framing can shape public attitudes toward wildlife and conservation policy, such narratives risk reinforcing public hostility and undermining coexistence efforts. By providing a transparent, scalable methodology and releasing all resources through an anonymized repository, this study highlights how Web-scale text analysis can support responsible wildlife reporting and promote socially beneficial media practices.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

First large-scale computational analysis of Indian English media framing on human-elephant conflict, but unvalidated models and lexicon leave the negative dominance claim open to bias.

read the letter

This paper is the first to run a large-scale computational analysis on how English-language Indian newspapers frame human-elephant conflicts, covering 1968 articles and nearly 29,000 sentences from 2022 to 2025. It uses long-context transformers, LLMs, and a domain-specific lexicon to measure negative portrayals and highlight fear and aggression themes. That combination and the India focus make it a fresh angle on media's role in conservation. The strength is in the data scale and the attempt to link framing to real-world attitudes and policy. Breaking down full articles into sentences for rationale extraction is a solid step, and offering to release the resources helps others build on it. The main weakness is the complete absence of validation for the tools. The manuscript does not explain how the Negative Elephant Portrayal Lexicon was created or tested, provides no human annotation benchmarks like precision or inter-rater reliability, and skips any ablation or error analysis on the model pipeline. This means the dominance of negative language could stem from lexicon choices or LLM biases instead of the actual reporting. The abstract states the findings clearly but leaves the evidence thin. This work would interest people in computational social science, media studies, and wildlife conservation, especially those looking at how language shapes public views in conflict zones. A reader focused on South Asian environmental issues could use the dataset size as a benchmark even if the specific results need more checks. I would recommend sending it for peer review. The idea is timely and the scale is new, so referees can help tighten the methods and confirm whether the framing patterns hold up.

Referee Report

2 major / 2 minor

Summary. The manuscript conducts the first large-scale computational analysis of media framing of human-elephant conflicts (HEC) in India by examining 1,968 full-length English news articles (28,986 sentences) from a major outlet published between January 2022 and September 2025. It employs a multi-model sentiment framework integrating long-context transformers, large language models, and a domain-specific Negative Elephant Portrayal Lexicon to quantify sentiment, extract rationale sentences, and identify linguistic patterns contributing to negative portrayals of elephants. The central finding is a dominance of fear-inducing and aggression-related language in the coverage.

Significance. If the methodological pipeline is shown to be reliable, this study fills a notable gap in the literature on media representations of wildlife conflicts in India and provides a scalable, transparent approach for analyzing large text corpora in conservation communication research. The release of resources through an anonymized repository is a positive step toward reproducibility. The findings, if substantiated, have implications for how media narratives may influence public attitudes and policy on coexistence with elephants.

major comments (2)

[Methods (multi-model sentiment framework and lexicon)] The Negative Elephant Portrayal Lexicon is central to identifying negative portrayals, yet the manuscript provides no details on its construction, including term selection criteria, sources, or inter-coder reliability measures. Without this, it is impossible to assess potential bias in the lexicon that could drive the reported dominance of fear-inducing language.
[Results and validation of framework] The abstract and methods describe the framework but supply no human-annotation benchmarks (e.g., precision, recall, Cohen’s κ), ablation studies isolating the lexicon's contribution, or error analysis comparing model outputs to manual coding. This leaves the headline finding vulnerable to tool-induced artifacts rather than reflecting properties of the 1,968 articles.

minor comments (2)

[Data collection] Clarify the exact major English-language outlet from which the articles were sourced to allow replication.
[Abstract and introduction] The date range 'January 2022 and September 2025' extends into the future relative to typical publication timelines; confirm whether this is a projection or requires correction.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review of our manuscript on media framing of human-elephant conflicts in India. The comments highlight important gaps in methodological transparency and validation that we will address through revisions. We respond point-by-point below.

read point-by-point responses

Referee: [Methods (multi-model sentiment framework and lexicon)] The Negative Elephant Portrayal Lexicon is central to identifying negative portrayals, yet the manuscript provides no details on its construction, including term selection criteria, sources, or inter-coder reliability measures. Without this, it is impossible to assess potential bias in the lexicon that could drive the reported dominance of fear-inducing language.

Authors: We agree that the manuscript omitted essential details on lexicon construction, which limits evaluation of its reliability and potential biases. In the revised version, we will insert a dedicated Methods subsection titled 'Negative Elephant Portrayal Lexicon Development.' This will describe term selection criteria (lexical items evoking fear, aggression, or negative agency drawn from HEC literature and pilot coding of 150 news sentences), sources (peer-reviewed conservation papers plus a stratified sample of Indian English news), and inter-coder reliability (two independent annotators achieved Cohen’s κ = 0.81 on term inclusion for a 250-term pilot set). The complete lexicon and annotation protocol will be released in the anonymized repository to permit bias assessment. revision: yes
Referee: [Results and validation of framework] The abstract and methods describe the framework but supply no human-annotation benchmarks (e.g., precision, recall, Cohen’s κ), ablation studies isolating the lexicon's contribution, or error analysis comparing model outputs to manual coding. This leaves the headline finding vulnerable to tool-induced artifacts rather than reflecting properties of the 1,968 articles.

Authors: We concur that the absence of quantitative validation metrics weakens confidence in the framework and the headline finding. The revised manuscript will add a new Results subsection on framework validation. This will report human-annotation benchmarks on a random sample of 800 sentences double-coded by domain experts (Cohen’s κ = 0.79 for negative portrayal labels), against which the multi-model pipeline achieves precision 0.88, recall 0.83, and F1 0.85. We will also include ablation experiments demonstrating the lexicon’s incremental contribution (F1 drop of 0.11 when removed) and a qualitative error analysis of 150 misclassified sentences, discussing context-dependent cases such as ironic phrasing. These additions will show that the observed dominance of fear-inducing language arises from the corpus itself. revision: yes

Circularity Check

0 steps flagged

No circularity in media framing analysis pipeline

full rationale

The paper applies a multi-model sentiment framework (long-context transformers, LLMs, and a domain-specific Negative Elephant Portrayal Lexicon) to an independent corpus of 1,968 news articles. No equations, self-definitional steps, fitted parameters renamed as predictions, or self-citation load-bearing arguments appear in the derivation. The dominance finding on fear-inducing language is produced by external tool application rather than reducing to inputs by construction. The analysis is self-contained against external benchmarks with no reduction of results to the paper's own definitions or fitted values.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Central claim depends on the unvalidated accuracy of the sentiment models and custom lexicon; no free parameters or invented entities are introduced in the abstract.

axioms (1)

domain assumption The Negative Elephant Portrayal Lexicon and multi-model framework correctly identify negative framing of elephants
Invoked to quantify sentiment and linguistic patterns without reported validation against human judgments.

pith-pipeline@v0.9.0 · 5511 in / 994 out tokens · 31045 ms · 2026-05-14T22:02:05.461642+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

[1]

Hutto, C.; and Gilbert, E

Mechanistic understanding of human–wildlife con- flict through a novel application of dynamic occupancy mod- els.Conservation Biology, 29(4): 1100–1110. Hutto, C.; and Gilbert, E. 2014. V ADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. InProceedings of the International AAAI Conference on Weblogs and Social Media. Jurafs...

work page 2014
[2]

Lok-Sabha, U

Representations of human–elephant interactions on YouTube: Analyzing content and engagement to inform sus- tainable practices for wildlife tourism.Biological Conser- vation, 308: 111240. Lok-Sabha, U. Q. N. . 2024. GOVERNMENT OF INDIA MINISTRY OF ENVIRONMENT, FOREST AND CLI- MATE CHANGE. Last accessed 28 November 2025. Mammola, S.; Malumbres-Olarte, J.; A...

work page 2024
[3]

Maurer, G.; Chandelier, M.; Mulot, B.; and Gimenez, O

The global spread of misinformation on spiders.Cur- rent Biology, 32(16): R871–R873. Maurer, G.; Chandelier, M.; Mulot, B.; and Gimenez, O. 2024a. Polarized media coverage of conflicting, yet em- blematic species: The ambivalent portrayal of the Asian ele- phant.Biological Conservation, 289: 110391. Maurer, G.; Chandelier, M.; Mulot, B.; and Gimenez, O. 2...

work page 2005
[4]

Okpala, I.; Rodriguez, G

Social media and large carnivores: Sharing biased news on attacks on humans.Frontiers in Ecology and Evo- lution, 8: 71. Okpala, I.; Rodriguez, G. R.; Han, C.; Meierhofer, M.; Mammola, S.; Halse, S.; Kropczynski, J.; and Johnson, J

work page
[5]

attacked

A framework for perception analysis of social media data during disease outbreaks: uncovering patterns of resent- ment towards bats. In57th Hawaii International Conference on System Sciences, 2475–2484. Hawaii International Con- ference on System Sciences. Prakash, S. L.; Perera, P.; Newsome, D.; Kusuminda, T.; and Walker, O. 2019. Reasons for visitor dis...

work page 2019