Language, agency, and deliberation: a multilevel theoretical framework for the computational analysis of citizen participation

Abstract

Computational analysis of citizen participation discourse has developed a substantial toolkit without a corresponding account of which linguistic dimensions carry psychological and political information. This article proposes a multilevel framework to fill that gap and states its claims as hypotheses rather than findings. Three analytical dimensions are identified: content, which covers lexical and topical features of what citizens say; interactional synchrony, which covers Language Style Matching (LSM) and the coordination of function words between participants; and narrative positioning, which aggregates semantic role profiles across a corpus to characterize the stance from which citizens narrate their relation to institutions. These dimensions are treated as complementary rather than as an ordered progression. We derive the framework from a decade of work with Chilean participatory corpora and place it in dialogue with deliberative theory, arguing that low interactional synchrony counts as evidence against communicatively oriented exchange while high synchrony does not count as evidence for it. This asymmetry turns the framework into an instrument for detecting deliberative failure.

The authors' abstract, as published at the source. Frontiers in Artificial Intelligence, 2026 · DOI ↗