Leadership discourses and organizational citizenship behavior in digital publics: a computational analysis of LinkedIn comment narratives

Introduction Despite extensive research on organizational citizenship behavior (OCB) and leadership, most studies rely on survey-based self-report methods that are susceptible to social desirability bias and common method variance. Less is known about how OCB-related orientations are discursively signaled in naturally occurring professional digital communication. This study examines the linguistic and thematic construction of OCB-related discourse within leadership-related LinkedIn narratives. LinkedIn comments are treated as discursive artifacts rather than direct evidence of actual citizenship behavior. Methods The study analyzed 12,421 user-generated LinkedIn comments collected through keyword-based sampling. A multi-method computational text analysis framework was employed, integrating BERTopic modeling, transformer-based sentiment analysis, and L2-regularized logistic regression. OCB-related language was operationalized through a validated dictionary based on the canonical dimensions of OCB, while discourse themes, leadership keyword presence, sentiment, and comment length were examined as predictive variables. Results OCB-related discourse accounted for 32.7% of the corpus and was unevenly distributed across thematic clusters. Collaborative leadership and knowledge-sharing discourse exhibited the highest OCB-related prevalence at 84%, followed by mentorship and ethical leadership themes. In contrast, abstract leadership praise, generic endorsement, and non-relational commentary contained virtually no OCB-related linguistic signals. Leadership keyword presence showed a negative predictive association with OCB-related discourse ( β = −0.77), whereas collaborative leadership discourse showed the strongest positive association ( β = 2.36; OR = 10.59). Sentiment differences between OCB-related and non-OCB comments were statistically significant but substantively negligible. Discussion The findings indicate that citizenship-oriented language in digital professional environments is shaped more strongly by relational, collaborative, and trust-based discourse contexts than by surface-level leadership terminology. The study contributes an exploratory computational framework for identifying OCB-related discursive signals in digital trace data and uses Social Exchange Theory as an interpretive lens for understanding how relational exchange cues may appear in professional online communication.