Identifying Digital Innovation Through Discursive Cohesion: A Statement-Based Computational Approach
Paper Number
ECIS2026-1214
Paper Type
CRP
Abstract
Digital innovation research provides valuable insights into how firms should organize, govern, and strategize around digital technologies. However, identifying where digital innovation materializes within organizations remains challenging. Using AI to analyze organizational language offers a promising way forward, as innovation emerges through employee discourse. Yet standard computational methods face a limitation: embedding-based models trained on public text cannot capture specialized jargon signaling organizational practices. We develop and validate a discourse-based approach to identifying digital innovation initiatives (DIIs) through discursive cohesion. This concept describes how employees develop specialized language to coordinate action around digital technologies. We computationally operationalize Foucault’s statements by isolating organizational-specific jargon and using it as clustering features. Analyzing employee profiles at a European automaker, we surfaced two DIIs: one visible collaboration within organizational structure and one latent collaboration dispersed across boundaries. The latent DII demonstrates that discursive cohesion characterizes employee-driven innovation even when formal organizational ties are absent.
Recommended Citation
Svahn, Fredrik and Øvrelid, Egil, "Identifying Digital Innovation Through Discursive Cohesion: A Statement-Based Computational Approach" (2026). ECIS 2026 Proceedings. 1.
https://aisel.aisnet.org/ecis2026/bpm/bpm/1
Identifying Digital Innovation Through Discursive Cohesion: A Statement-Based Computational Approach
Digital innovation research provides valuable insights into how firms should organize, govern, and strategize around digital technologies. However, identifying where digital innovation materializes within organizations remains challenging. Using AI to analyze organizational language offers a promising way forward, as innovation emerges through employee discourse. Yet standard computational methods face a limitation: embedding-based models trained on public text cannot capture specialized jargon signaling organizational practices. We develop and validate a discourse-based approach to identifying digital innovation initiatives (DIIs) through discursive cohesion. This concept describes how employees develop specialized language to coordinate action around digital technologies. We computationally operationalize Foucault’s statements by isolating organizational-specific jargon and using it as clustering features. Analyzing employee profiles at a European automaker, we surfaced two DIIs: one visible collaboration within organizational structure and one latent collaboration dispersed across boundaries. The latent DII demonstrates that discursive cohesion characterizes employee-driven innovation even when formal organizational ties are absent.