Online opinions, sentiments and news framing of the first nuclear referendum in Taiwan: a mix-method approach

ABSTRACT

This mixed-method study uses a big data approach to examine cross-platform public sentiments towards Taiwan’s first nuclear energy referendum, and further conducts content analysis for nuclear news framing strategies. Sentiment analysis shows polarized affective attitudes towards Go Green with Nuclear (GGWN) referendum, regardless of media types. News coverage and social media contents reveal significant sentiment differences in narrating the referendum, nuclear energy, and political party-related issues. Polarized political party-related nuclear claims tend to show negative sentiments. As for agenda setting, the big data analysis shows that politics dominate nuclear narratives on news, Facebook and forums. In addition, content analysis reveals that the majority of news articles involve politics, but rarely report on energy and environmental subjects. In terms of generic news framing strategies, dramatic framing is used more than substantive framing in nuclear narratives. Conflict is the leading framing, followed by action. As for environmental news framing, most GGWN-related news is not eco-centric. Eco-efficient framing is most used to emphasize economic growth, national development and people’s livelihood. Moreover, mainstream and alternative media show no significant differences in using generic and environmental news framing to report nuclear referendum issues. Implications are discussed.

KEYWORDS:

• Nuclear energy policy
• sentiment analysis
• polarization
• generic news framing
• environmental news framing

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 uMiner platform is developed by uMaxDATA to generate insights from monitoring and mining massive unstructured online data. It uses social listening and artificial intelligence machine learning technologies for big data intelligence mining and media coverage collection.

2 SAM calculates filtered subjects’ collective affective tendency based on accumulated daily content and emotional trends. The frequency of each sentiment value is computed by time series on the uMiner platform.

Additional information

Notes on contributors

Trisha T. C. Lin

Trisha T.C. Lin (Ph.D., University of Hawaii, Manoa) is professor at Department of Radio and Television, College of Communication, National Chengchi University, Taiwan. She is the research fellow of Taiwan Institute for Governance and Communication Research. In addition to the former broadcasting media professional, she worked as the assistant professor at Nanyang Technological University, Singapore. Her research interests focus on examining emerging interactive digital media with two approaches: socio-technical system analysis and socio-psychological user research. She published nearly 60 new media-related journal articles regarding mobile media and communication, adoption and use of emerging media technologies, digital journalism, and health communication.