Tweet Emotion Dynamics: Emotion Word Usage in Tweets from US and Canada

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 4162–4176 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 4162 Tweet Emotion Dynamics: Emotion Word Usage in Tweets from US and Canada Krishnapriya Vishnubhotla∗‡ and Saif M. Mohammad† ∗Department of Computer Science, University of Toronto ‡Vector Institute for Artificial Intelligence vkpriya@cs.toronto.edu †National Research Council Canada saif.mohammad@nrc-cnrc.gc.ca Abstract Over the last decade, Twitter has emerged as one of the most influential forums for social, political, and health discourse. In this paper, we introduce a massive dataset of more than 45 million geo-located tweets posted between 2015 and 2021 from US and Canada (TUSC), especially curated for natural language analysis. We also introduce Tweet Emotion Dynamics (TED) — metrics to capture patterns of emotions associated with tweets over time. We use TED and TUSC to explore the use of emotion-associated words across US and Canada; across 2019 (pre-pandemic), 2020 (the year the pandemic hit), and 2021 (the second year of the pandemic); and across individual tweeters. We show that Canadian tweets tend to have higher valence, lower arousal, and higher dominance than the US tweets. Further, we show that the COVID-19 pandemic had a marked impact on the emotional signature of tweets posted in 2020, when compared to the adjoining years. Finally, we determine metrics of TED for 170,000 tweeters to benchmark characteristics of TED metrics at an aggregate level. TUSC and the metrics for TED will enable a wide variety of research on studying how we use language to express ourselves, persuade, communicate, and influence, with particularly promising applications in public health, affective science, social science, and psychology. Keywords: Emotions, Sentiment analysis, Tweets, Valence, Arousal, Dominance, Corpus Linguistics 1. Introduction Over the last decade, Twitter has emerged not only as one of the most influential micro-blogging platforms, but also one of the most actively engaging (if some- times polarizing) fronts for social, political, and even health discourse. Early work (Pak and Paroubek, 2010; Dodds et al., 2011) identified tweets as a crucial indica- tor of public sentiment. Since then, various samples of tweet data have been used to analyze a wide variety of phenomena, including the recent COVID-19 pan- demic. However, past work largely uses topic-based keywords to obtain datasets of interest (often at the ex- pense of geo-location information); for example, work that analyzes emotions in tweets that mention COVID- 19-associated terms (Banda et al., 2020; Lwin et al., 2020). Further, very little work explores changes in patterns of emotions of individuals over time. This paper introduces a new framework to analyze patterns of emotions associated with tweets over time, which we refer to as Tweet Emotion Dynamics (TED). TED builds on ideas first introduced in Hipson and Mohammad (2021), and applies metrics such as home base, variability, and rise rate to tweets. We also in- troduce a new dataset of geo-located English Tweets from US and Canada (TUSC). TUSC is not restricted to specific topics and so can be used to …

Content truncated to 500 words