Profiles and temporal dynamics of personality-related language in a 16-year professional email archive: a subject-specific study (2008-2024).

Abstract

This article reports a subject-specific, within-person self-study of personality-related language in one participant-author's predominantly professional email archive (2008-2024; n = 25,780). The aim is not between-person prediction or human-parity benchmarking, but idiographic description of one person's profile and temporal dynamics across Self-Determination Theory, the Big Five, and Psychological Well-Being. Email texts were coded with two LLM-assisted workflows – prompted GPT-3.5 ratings and a GPT-4.1 teacher distilled into compact multilingual transformer students – with cross-model comparisons treated as robustness checks. Email-level labels were aggregated into monthly balance indices on a common [-100, 100] scale. Within-person profile correspondence with contemporaneous self-report reference profiles was examined on POMP-aligned scales, and monthly shifts across pre-specified major personal and occupational periods were estimated using HAC-robust regressions with FDR control. Across frameworks, the archive yielded interpretable subject-specific patterns: email-derived profiles partly approximated the self-report reference shape while also reflecting the professional register, and month-level indices shifted in theory-coherent ways across salient periods. The findings support de-identified digital traces as state-sensitive adjuncts for person-oriented research on one individual's functioning over time. The results do not warrant between-person inference, and self-reports remain the primary reference for trait baselines.