Locally Ensconced and Globally Integrated: How Network Cohesion and Range Relate to a Language-Based Model of Group Identification

Details

Co-Authors

Douglas Guilbeault, Austin van Loon, Katharina Lix, Amir Goldberg, and Sameer B. Srivastava

Category of Paper

Peer-Reviewed Research Papers

Tags

AI, Computational Linguistics, Field Study, Machine Learning, Natural Language Processing, Network Cohesion, Network Range, Organizational Identification, Pronoun Use, Social Networks

Abstract

Shifting attachments to organizations are a constant in the modern era. What accounts for variation in the strength of organizational identification? Whereas prior work has emphasized organization-level properties and individual differences, this article instead highlights the role of network-structural positions. Distilling insights from prior work on networks and identity, the authors propose that organizational identification strength is positively related to network cohesion—having contacts who are mutually interconnected. Departing from prevailing accounts, they further propose that identification strength also varies via network range—having contacts who inhabit a broad range of network communities. Using the tools of computational linguistics to develop a language-based measure of identification, they find support for the theory using pooled data of internal communications from three organizations.

Locally Ensconced and Globally Integrated: How Network Cohesion and Range Relate to a Language-Based Model of Group Identification.” American Journal of Sociology 131(1): 149-199. [Data and Code]