Catching up with trends: The changing landscape of political discussions on twitter in 2014 and 2019

Avinash Tulasi, Kanay Gupta, Omkar Gurjar, Sathvik Sanjeev Buggana, Paras Mehan, Arun Balaji Buduru, Ponnurangam KumaraguruarXiv 2019 · arXiv

Abstract

This study provides a longitudinal analysis of Twitter's transformation from an elite, English-centric platform into a massive, multilingual political battleground during the 2014 and 2019 Indian General Elections. By analyzing millions of tweets, researchers from IIIT Delhi and IIIT Hyderabad quantify a 600% increase in active politician handles and a significant linguistic pivot toward Hindi and regional languages.

Introduction

The digital landscape in India underwent a seismic shift between 2014 and 2019, largely driven by the "Jio effect"—the arrival of affordable 4G data. This paper examines how political parties adapted their strategies to this new, broader user base, moving from elite signaling to mass-market engagement.

Methodology

The researchers compared two massive datasets: 18.7 million tweets from 2014 and 45.1 million from 2019. To analyze this data, they introduced an "Efficiency" metric ($e_i$), defined as the ratio of total retweets to total tweets made by a handle. This metric identifies which politicians get the most "bang for their buck" in terms of engagement. Linguistic analysis using Z-scores and manual topic classification were also employed to identify thematic priorities and strategic focuses.

The Mobile and Vernacular Pivot

One of the most striking findings is the shift in how and in what language users engage. In 2014, 91% of political tweets were in English. By 2019, Hindi became the dominant language (53.4%), with significant growth in regional languages like Tamil and Marathi.

Simultaneously, the source of these tweets shifted dramatically toward mobile platforms. In 2014, web-based clients were prominent; by 2019, mobile platforms (Android and iPhone) accounted for over 78% of content, reflecting the explosion of smartphone penetration in India.

Shift in tweet sources from 2014 to 2019, showing the rise of Android dominance.

Strategic Divergence and Efficiency

The study highlights a clear difference in campaigning styles between the two major parties. Topic modeling revealed that Narendra Modi’s most frequent topic was "Campaign/Plan/Manifesto" (23%), focusing on proactive achievement-oriented messaging. Conversely, Rahul Gandhi’s most frequent topic was "Anti-Modi" (37.2%), suggesting a more reactive and monotonous strategy.

The "Efficiency" metric further revealed that while individual opposition leaders might have high engagement, the BJP as a whole maintained a higher probability of high-efficiency handles across their broader candidate list compared to the INC.

Probability distribution of handle efficiency for INC and BJP, showing the BJP's higher probability of high-efficiency handles.

Key Results and Reach

  • Scale: Twitter users in India grew from 15.8 million in 2014 to 34.4 million in 2019.
  • Reach: Narendra Modi’s follower count saw a 20x increase (from ~2.4M to ~47M).
  • Engagement: The average number of retweets per politician tweet jumped by 114%.

Conclusion

The evolution of Twitter in India reflects broader societal changes in digital access. The transition from an elite tool to a mass-market political weapon was characterized by linguistic diversification, mobile-first strategies, and a shift toward high-efficiency, proactive messaging. Future research is needed to address the impact of automation and misinformation in this expanded digital arena.

Summary updated

BibTeX
@misc{tulasi2019catching,
  author = {Avinash Tulasi and Kanay Gupta and Omkar Gurjar and Sathvik Sanjeev Buggana and Paras Mehan and Arun Balaji Buduru and Ponnurangam Kumaraguru},
  title = {{Catching up with trends: The changing landscape of political discussions on twitter in 2014 and 2019}},
  year = {2019},
  eprint = {1909.07144},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1909.07144},
}
Figure