Attitudes Towards Global Warming on Twitter: A Hedonometer-Appraisal Analysis

Introduction

As one of the major concerns worldwide, global warming and its consequences have prompted exhaustive discussion in academic literature and the media, from traditional media such as newspapers to new media such as Twitter and Facebook. As the media plays a critical role in shaping the public’s views by driving and perpetuating concepts and opinions, many studies focus on global warming reports in newspapers from the perspective of topical prevalence over time (Bohr, 2020), linguistic patterns in global warming reports (Dayrell, 2019), or influence of news reports on public understanding (Jang, 2013). However, the arrival of the digital age and the rise of social media have attracted academic attentions from traditional forms of unidirectional communication such as the print press or television to new ways of bidirectional interaction such as Facebook, YouTube and Twitter. Besides, the previous age of mass communication disseminating information to the public has evolved into self-communication which communicates with oneself. This results in media discourse transforming from merely “a unified, generic ‘hypertext’” to a “diversified, individualised ‘mytext’” (Castells, 2013, p. xx). Social media allows users to share their opinions and attitudes alongside various sources that contribute to the public’s response to current issues. Increasing acknowledgement of the significance of the public’s perception of global warming, which is essential for the formation of public attitudes to this issue and changes of social practices as well as policymaking, has led to more scholarship on social media discourse. Use of sentiment analysis on global warming is now one of the leading ways to ascertain the general polarity of a given text. However, mere awareness of this polarity is not enough for discourse analysis, which requires a greater level of specificity, such as the linguistic representation of global warming in social media. Besides, subjective involvement and use of relatively small data sets, and the “cherry-picking” phenomenon (only choosing supporting content as evidence to support a point), are often criticised in the literature of linguistics (Baker et al., 2008). Thus far, insufficient scholarly attention has been paid to the linguistic representation of global warming in terms of attitudinal statements while considering the general polarity of a large amount of data.

In order to fill this gap in the literature, the present study explores messages concerning global warming on Twitter, one of the most popular social networking platforms, by applying big data analytics for discourse analytic purposes and proposing a new framework: the combination of Hedonometer and Appraisal Theory. The sentiment analysis technique Hedonometer is responsible for the general tendency toward global warming on Twitter and Appraisal Theory for scrutiny of linguistic realisation. By applying a text-mining technique and traditional discourse analysis method, it is possible to meet both demands of grasping details of linguistic representation and general polarity of considerable data avoiding manually laborious work as well as “cherry-picking” phenomenon.

The paper makes the following main contributions. First, the research shows the possibility of combining sentiment analysis technique and linguistic theory. Specifically, the newly proposed framework constructs a complementary role for each other by combining Hedonometer and Appraisal Theory. As mentioned earlier, the framework maximizes benefits and minimizes disadvantages of both. Second, the research demonstrates the general attitudes towards global warming on social media and how the attitudes are conveyed in language, which is, heretofore, the first attempt in the literature of attitude analysis. Third, the research contributes to the sociological and psychological studies by confirming the negative bias phenomenon and analysing different perspectives taken in the use of person, respectively. The present paper is structured into two procedures. The first procedure provides an overview of public opinions toward global warming on Twitter between January 2020 and April 2021. It employs Hedonometer to indicate attitudinal polarity, and word shift graphs to analyse specific words contributing to the polarities and how they do so. The second procedure scrutinises how public attitudes are presented on Twitter using Appraisal Theory, examining linguistic features in detail.