Background Masking Alters Accuracy and Confusion Structure in Mandarin Emotional Prosody Perception

Emotional prosody is a key cue for affective communication, yet everyday speech often occurs in background noise. It remains unclear whether masking only reduces Mandarin emotional prosody recognition accuracy or also alters emotion-confusion structure. We examined this issue with speech-shaped noise (SSN) and multi-talker babble (MTB). Thirty-six native Mandarin-speaking young adults identified semantically neutral sentences conveying happiness, sadness, anger, and surprise in silence and under SSN and MTB at 0 dB and -4 dB signal-to-noise ratios. Recognition accuracy was analyzed using trial-level generalized linear mixed-effects models, whereas confusion structure was assessed using confusion matrices, sadness-high-arousal boundary errors, and Net Flow, an index of directed target-response asymmetry. Accuracy declined under masking relative to silence, with the poorest performance in the MTB -4 dB condition. Across emotions, estimated accuracy was lowest for surprise and comparatively high for sadness. The MTB -4 dB condition increased sadness-high-arousal boundary errors; Net Flow characterized a source-like pattern for surprise and a more sink-like pattern for sadness, particularly in this condition. These findings suggest that background masking affects Mandarin emotional prosody perception not only by reducing recognition accuracy but also by altering directed response-error structure.