From 14 thematic categories manually annotated from 1,000 keywords by two blinded linguists, contrasted against the enTenTen15 reference corpus.
Real LIWC-2015 scores from the paper across all 14 weeks. Analytical thinking remained stably high even as emotional tone fluctuated with the pandemic's trajectory.
English-speaking tweeters maintained high Analytic scores (81–93) throughout — indicating logical, hierarchical thinking, not narrative panic. Clout spiked at W8 (80.02) when social solidarity peaked, then declined. Tone remained consistently below 50 (negative) for 12 of 14 weeks.
Anxiety peaked early at W3 (0.66) then paradoxically fell as cases exploded. Anger kept climbing to 0.87 at W10–W11 as institutional failures mounted. Sadness rose steadily to 0.51 by W13.
Using MetaNet's 684-metaphor database and manual concordance annotation, 8 dominant conceptual metaphors structured how English-speaking tweeters understood COVID-19. Click to explore.
Work-related language grew while leisure collapsed. Religion doubled at W8 as lockdowns began. Cognitive processing spiked at W10–W11 when rationalizing a collapsing world became urgent.
Leisure language crashed from 2.25 (W1) to 0.76 (W14) — a 66% decline. This linguistic shift mirrors a behavioral transformation: as the pandemic became the new default reality, casual and recreational discourse was replaced by health, work, and crisis vocabulary.
All four summary variables and key emotion scores from Tables 2–3 of the paper. Rows highlighted on hover.