When Faster Backfires: Affective Primacy and Speech Rate in Live-Streamed Knowledge Payment Services

Live-streamed online knowledge payment (OKP), where creators sell expertise in real time through speech, is a major digital-service market, yet creators lack evidence on how fast to speak. Using the Elaboration Likelihood Model (ELM) as an interpretive lens, we examine associations among speech rate, two behavioral engagement indices (attentional engagement, measured by viewer dwell time, and affective engagement, measured by likes and emoji), and purchase-link clicks in 526 no-presenter Douyin knowledge sessions, with speech rate measured by automated speech recognition. Two findings stand out. First, both engagement indices have an inverted-U association with speech rate, strongest at intermediate rates, and formal endpoint tests reject a uniformly positive relationship (p < 0.001), although the curvature weakens once the sparse fastest sessions are removed. Second, the affective index has a standardized association with purchase-link clicks about 2.5 times that of the attentional index (difference = 0.180, 95% CI [0.071, 0.287]), a behavioral pattern we call affective primacy. Because the design is observational, we present the ELM reading and the role of transactional time pressure as interpretation rather than tested mechanism, and we note that the turning points are imprecise and sensitive to the rare fastest sessions. The study contributes calibrated, reproducible evidence that speaking faster is not uniformly better and that real-time affective response tracks clicking more closely than sustained attention does.