Does the Appearance of Autonomous Conversational Robots Affect User Spoken Behaviors in Real-World Conference Interactions?

2025-03-17

AI Summary

This study investigates how robot appearance influences users’ spoken language behavior during a real academic conference, specifically comparing the highly anthropomorphic android ERI CA with the less anthropomorphic telepresence robot TELECO. Method: Based on transcribed speech from 42 attendees, we quantified disfluency rates and syntactic complexity using computational linguistics methods, and applied Naïve Bayes classification, feature importance analysis, and effect-size estimation (Cohen’s *d*). Contribution/Results: For the first time in an authentic conference setting, we empirically demonstrate that high anthropomorphism (ERI CA) significantly reduces disfluency and increases syntactic complexity—both with medium effect sizes—supporting dual mechanisms of reduced cognitive load and enhanced verbal adaptation. The classifier achieved 71.60% F1-score, confirming disfluency and syntactic complexity as key discriminative linguistic features for robot appearance. These findings provide interpretable, linguistically grounded evidence to inform embodied interaction design.

Technology Category

• Intelligent Robots: Human-Robot Interaction
• Natural Language Processing: Safety and Robustness
• Humans and AI: Interaction Techniques and Devices

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

We investigate the impact of robot appearance on users' spoken behavior during real-world interactions by comparing a human-like android, ERICA, with a less anthropomorphic humanoid, TELECO. Analyzing data from 42 participants at SIGDIAL 2024, we extracted linguistic features such as disfluencies and syntactic complexity from conversation transcripts. The results showed moderate effect sizes, suggesting that participants produced fewer disfluencies and employed more complex syntax when interacting with ERICA. Further analysis involving training classification models like Na""ive Bayes, which achieved an F1-score of 71.60%, and conducting feature importance analysis, highlighted the significant role of disfluencies and syntactic complexity in interactions with robots of varying human-like appearances. Discussing these findings within the frameworks of cognitive load and Communication Accommodation Theory, we conclude that designing robots to elicit more structured and fluent user speech can enhance their communicative alignment with humans.

Problem

• Impact of robot appearance on user spoken behavior
• Comparison of human-like android vs. less anthropomorphic robot
• Role of disfluencies and syntactic complexity in interactions

Innovation

• Compare human-like android ERICA with less anthropomorphic TELECO
• Analyze linguistic features like disfluencies and syntactic complexity
• Use Naïve Bayes for classification and feature importance analysis