Linguistic features influencing information diffusion in social networks: A systematic review

Abstract

This review summarizes research empirically addressing the link between linguistic features and spread of information in online social media. N = 85 studies were selected for the review. Features that consistently had a positive effect on diffusion were information value, words pertaining to social connection, as well as engaging, concrete and powerful language style. Instead, effects of sentiment polarity (positive vs. negative), basic emotions, and morality vary in their relation with diffusion. We also noted whether the diffusion has been investigated as influence- or homophily-based, i.e., whether the sharing occurred as the result of the message exposure or simply because of inherent similarity between the users. This difference in sharing was considered in only n = 7 of the selected works, of which only n = 2 identified features that positively affect both processes.