Does negatively toned language use on social media lead to attitude polarization?

Prior research has indicated that both attitudinal homogeneity of communication networks (“echo chambers”) and heterogeneity (“adversarial debates”) can lead to attitude polarization. The present paper argues in echo chambers adversarial debates is dominated by network negativity, a negative valence the tone discussions which might be associated with Combining methods from sentiment analysis social analysis, more than 4 million tweets on two controversial topics (Brexit, Trump) were analyzed investigate occurrence negativity its association proxies polarization (extremity ambivalence). Results indicate users' own was most strongly related polarization, whereas among users’ friends, or consonance sentiments between users friends had less impact findings are literatures bias, optimal distinctiveness theory, intergroup contact theory.