This article investigates the language of happiness from two perspectives: the role of sentiment words in self-reported descriptions and the identification of happiness sources. Analyzing the HappyDB corpus using text analytics, the study finds that positive lexical items play a limited role. Unsupervised machine learning extracts and clusters keywords, revealing semantic classes describing happiness sources. Named entity analysis highlights the importance of commercial products and services. The research provides methodological underpinnings for processing self-reported happy moments and contributes to understanding the linguistic expression of happiness.

Antonio Moreno-Ortiz, Chantal Pérez-Hernández, María García-Gámez