Too much of a good thing: nonlinear effect of streamers’ language arousal on viewer engagement in live streaming

Purpose In the growing field of live streaming, viewers’ interactive engagement is particularly critical. Although streamers frequently use language arousal to enhance engagement, its specific effects remain insufficiently understood, necessitating further investigation. Drawing on the emotion as social information (EASI) theory, the present study examines the impact of streamers’ language arousal in live streaming on viewer engagement and explores the moderating role of streamer type. Design/methodology/approach Using real-time data collected from 162 live streams on the Douyin platform, we develop a negative binomial model to examine the relationship between streamers’ language arousal and viewer engagement. Findings Our empirical findings suggest that this influence of streamers’ language arousal on viewer engagement – in both paid and free forms – exhibits a nonlinear, inverted U-shaped relationship. Additionally, compared to micro-streamers, macro-streamers attenuate the inverted U-shaped relationship between streamers’ language arousal and viewer engagement. These baseline results are further validated through a rich set of robustness tests. Originality/value By examining the effect of streamers’ language arousal on viewer engagement, this study enhances the understanding of viewer behavior in live-streaming contexts. It also extends the EASI theory by introducing language arousal as a key emotional cue and demonstrating its explanatory power in live streaming, thereby broadening the theory’s applicability to interactive marketing environments. This study provides practical insights for optimizing engagement strategies in the live-streaming industry.