Date of Award

5-2026

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

School of Computing

Committee Chair/Advisor

Dr. Matias Volonte

Committee Member

Dr. Eric Patterson

Committee Member

Dr. Kelly Richardson

Abstract

Virtual Humans (VHs) are interactive systems capable of engaging users through spoken dialogue and coordinated nonverbal behavior. These virtual humans are increasingly used in domains such as healthcare, rehabilitation, and skill training because they allow users to practice communication within structured conversational exchanges. In contexts where verbal production and conversational engagement are central, virtual humans offer a promising platform for repeated and guided practice.

This thesis investigates how different visual interaction modalities affect user engagement with a Virtual Human. A pilot study was conducted using a between-subjects experimental design comparing three presentation conditions: a traditional screen-based 2D interface, a navigable 2D–3D virtual environment displayed on a monitor, and a fully immersive virtual reality (VR) environment. Across all conditions, participants interacted with the same conversational agent and completed a storytelling task in which they described a sequence of images presented by the system. The conversational dialogue, prompts, and agent behavior were held constant across conditions so that only the visual presentation modality varied.

The study examined several aspects of the interaction experience, including conversational verbosity, perceived presence, system usability, social presence, and participants’ subjective impressions and affective responses. Behavioral measures were derived from automatically logged conversation transcripts, while subjective evaluations were collected through post-interaction questionnaires. Because the study was conducted as an exploratory pilot investigation with a limited sample size, the analysis focused on identifying preliminary descriptive patterns across the three modalities.

Results indicate that participants interacting in the immersive conditions produced more speech during the storytelling task compared to those in the traditional 2D interface. In contrast, subjective evaluations of presence, usability, and social presence were broadly comparable across conditions. Affective responses and overall impressions of the conversational agent also showed only modest variation between modalities.

The findings provide initial insight into how the visual context of a virtual human interaction may relate to conversational engagement and user perception. As a pilot study, this work contributes early empirical evidence that can inform the design of future conversational systems and larger-scale investigations. Crucially, this pilot serves as a foundation for the ultimate goal of developing a speech impediment training system, a conversational platform in which individuals with speech difficulties, such as stutter and reduced articulation, can practice spoken communication with a virtual human in a structured, supportive, and clinically deployable environment.

Author ORCID Identifier

https://orcid.org/0009-0003-4386-0203

Download

Available for download on Monday, May 31, 2027

Share

COinS