Visual Feedback Design for Voice Assistants: Applying Nonverbal Backchannel and Turn-Taking Cues

In human conversation, seamless turn-taking and backchanneling are fundamental for maintaining natural dialogue. These dynamics rely on nonverbal signals to regulate conversational flow and indicate attention. However, replicating this fluency remains challenging due to the lack of physical embodiment and nonverbal cues in Voice User Interfaces (VUIs). Addressing this, we conceptualized four distinct conversational cues—including backchannel and turn-taking cues—for geometric icon-based VUIs, designed visual representations, and evaluated their effects. Study 1 (N = 65) examined three backchannel patterns (regular, synchronized, randomized). Synchronized and randomized feedback significantly improved perceptions of the conversational flow and the voice assistant (e.g., perceived active listening, social presence, trust) over regular feedback, with minimal differences between the two conditions. Study 2 (N = 39) investigated turn-taking cues (unified-channel, separated-channel, predictive-channel). The predictive-channel, featuring separated icons with explicit signaling, outperformed other conditions. Together, these findings provide empirical evidence that visual nonverbal cues can enhance user experience in non-embodied interfaces.