Age-related shifts in temporal narrative content: A structural topic modeling analysis of a life-span corpus

Abstract Across the life span, people recall their past and envision their future, yet how the content of these narratives changes with age has rarely been examined at scale. We constructed a rare cross-sectional corpus of short essays in which participants ranging from elementary-school children to adults aged 60–80 years each wrote about a past event and an anticipated future event, and we modeled topic prevalence with a Structural Topic Model (STM) that incorporates age, gender, and temporal reference (past versus future) as covariates. The model yielded ten interpretable topics without manual coding. Five topics showed significant age × gender interactions in prevalence, and the focus of future narratives shifted with age from work and self-oriented life planning toward family, the next generation, and health. Future narratives were also consistently more abstract than past narratives. The approach provides a scalable framework for quantifying age-related change in temporal narrative without large-scale manual coding.