Big Changes Start With Small Talk: Twitter and Climate Change in Times of Coronavirus Pandemic.

Behavioural scientists have been studying public perceptions to understand how and why people behave the way they do towards climate change. In recent times, enormous changes to behaviour and people's interactions have been brought about by the worldwide coronavirus disease 2019 (COVID-19) pandemic, unexpectedly and indefinitely; some of which have environmental implications (e.g., travelling less). An innovative way to analyse public perceptions and behaviour is with the use of social media to understand the discourse around climate change. This paper focuses on assessing changes in social media discourse around actions for climate change mitigation over time during the global pandemic. Twitter data were collected at three different points during the pandemic: February (time 1), June (time 2), and October 2020 (time 3). By using machine learning techniques, including recurrent neural networks (RNN) and unsupervised learning Latent Dirichlet Allocation (LDA) topic modelling, we identified tweets mentioning actions to mitigate climate change. The findings identified topics related to "government actions," "environmental behaviours," "sustainable production," and "awareness," among others. We found an increase in tweets identified as "action tweets" relating to climate change for time 2 and time 3 compared with time 1. In addition, we found that the topic of energy seemed to be of relevance within the public's perceptions of actions for climate change mitigation; this did not seem to change over time. We found that the topic of "government actions" was present across all time points and may have been influenced by political events at time 1, and by COVID-19 discourse at times 2 and 3. Moreover, topic changes over time within Twitter indicated a pattern that may have reflected restrictions on mobility as these tended to focus on individual and private sphere behaviours rather than group and public sphere behaviours. Changes in topic patterns may also reflect an increase in salience of certain behaviours (e.g., shopping), which may have received increased attention due to lockdown restrictions. Considering restrictions and adaptability challenges people face in times of a global pandemic may help to identify how to support sustainable behaviour change and the likely persistence of these changes.