An asymmetric echo: tracing social sentiment through economic cycles in a large-scale corpus of American newspapers

Introduction. This study investigates the relationship between social sentiment and macroeconomic cycles in the United States from 1919–1937, a period of profound economic transformation. It aims to quantify and analyse public emotion dynamics across the economic expansion period, the downturn period, and the recovery period, seeking to identify patterns in how macrosocial and economic factors influence social sentiment. Method. A computational approach was employed, using natural language processing (NLP) and the LIWC lexicon to conduct sentiment analysis on a large-scale corpus of digitised historical newspapers from the Chronicling America database. Analysis. Quantitative sentiment tendencies were compared using analysis of variance (ANOVA) and independent samples t-tests to identify significant variations and correlations across the distinct economic periods. Conclusion(s). Findings reveal an asymmetric correlation: while positive sentiment is pro-cyclical, negative sentiment was counterintuitively suppressed during economic downturns. We attribute this paradox to a dual mechanism of individual psychological compensatory control and macro-level governmental expectation management. This research provides a novel framework for historical sentiment analysis, challenging conventional narratives about public responses to crises and offering insights for contemporary public opinion governance.