Skip to content
Open access

Don’t Just Fill the Silence: Exploring Conversational Filler Strategies in Embodied Virtual Agents

Aug 2026 · ACM Transactions on Applied Perception · Vol 23, pp. 1-23 · 0 citations · 62 references

TL;DR

It is suggested that conversational fillers serve two roles: as temporal signals that reduce perceived delay and as context-sensitive cues that shape social perception, with meaningful benefits emerging only when fillers are contextually aligned.

Abstract

Conversational fillers, such as short vocalizations (e.g., “uh,” “um”) or brief phrases that manage conversational flow, are increasingly used in embodied virtual agents. Yet, their effects on users’ perceptions remain insufficiently understood. Thus, we investigated how different conversational filler strategies shape users’ perceptions and experiences during interaction with an embodied virtual agent. We implemented a large language model (LLM)-driven conversational system that selects fillers from predefined sets and integrates them with generated responses via a prompt-engineered mechanism that maintains continuity between the filler and the response. In a within-group study ( \(N=24\) ), participants interacted with the agent across four conditions: filler-free (baseline), short, general, and context-specific fillers. Results showed that all filler types significantly improved perceived response time and reduced estimated response time relative to the filler-free condition. However, context-specific fillers provided additional benefits, including higher perceived filler quality, rapport, likability, willingness for future interaction, and lower uncanny valley ratings. Additional analyses showed that higher perceived filler quality and more favorable perceived response time were correlated with more positive perceptions of the agent and the interaction experience. These findings suggest that conversational fillers serve two roles: as temporal signals that reduce perceived delay and as context-sensitive cues that shape social perception, with meaningful benefits emerging only when fillers are contextually aligned.

Read PDF

Similar papers

Open access Sep 2026

Affect-aware conversational adaptation in mixed reality procedural tasks

A mixed-reality conversational system that integrates voice interaction, LLM-driven dialogue, and prosody-based affect-aware adaptation with a modular client–server architecture is presented, enabling emotion-related cues inferred from vocal prosody to be incorporated without interrupting conversational flow.

Andrea Antonio Cantone, Matteo Ercolino, M. Sebillo et al. · 0 citations
#artificial intelligence Preprint Oct 2026

DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents

Large language models (LLMs) deployed as conversational agents are increasingly expected to follow procedural logic. While factual hallucination has received extensive attention, a different failure mode has been overlooked: procedural hallucination, in which an agent produces individually fluent and factually plausibl...

Sergio Gastón Burdisso, Esaú Villatoro-Tello, Dairazalia Sánchez-Cortés et al. · 0 citations
Book Open access Nov 2026

Typing vs. Dialogue Trees: Do Players Actually Want Open-Ended LLM Conversations in Games?

Large Language Models (LLMs) offer the potential to replace static dialogue trees with open-ended, generative interactions in video games. However, the user experience of this transition remains under-researched. We present Project Bard, a custom 2D RPG that directly compares player interactions with LLM-based NPCs (fr...

Tom Tucek, R. Gambardella, Mohammadamin Sadeghizad · 0 citations
Review Open access Oct 2026

“Can you hear me, ChatGPT?”: interaction patterns, challenges, and learner perceptions in out-of-class EFL speaking practice

Investigating EFL learners’ interaction patterns with ChatGPT as an out-of-class language partner and examines how these patterns and learner perceptions evolve over time suggests greater interactional engagement and growing familiarity with AI-mediated speaking practice.

Wen-Zheng Huang, Hao Zhou · 0 citations
#artificial intelligence Preprint Oct 2026

DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models?

This work introduces DuplexAct, a benchmark for evaluating interaction action taking in E2E FD-SLMs and defines a three-level action taxonomy spanning eight interaction actions and 24 corresponding interaction conditions, and develops action-specific evaluation metrics for both.

Jia-Heng Dong, Shi-Hong Tan, Gong-Ping Huang et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.