Bachelor's-dissertation pilot of an early Samsung Gear VR public-speaking prototype with 6 adults who stutter - mixed anxiety results across participants

Background

Public speaking is widely reported as one of the most anxiety-provoking situations for people who stutter. Graduated exposure - repeatedly facing feared situations in a safe and structured way - is a well-established approach for reducing social anxiety, but arranging real audiences for repeated practice is impractical. By 2016, consumer VR headsets had become affordable enough (Samsung Gear VR / Oculus Mobile, ~£100 plus a compatible Samsung phone) to explore whether they could serve as a practical platform for self-directed speaking practice. This bachelor’s dissertation project at Nottingham Trent University set out to build and pilot-test such a prototype with adults who stutter.

What the researcher did

Walkom developed a custom VR application for the Samsung Gear VR headset (running on a Samsung S6 Android phone), placing the user at a podium in a virtual lecture hall. The hall contained three animated audience characters (not a ‘filled’ hall) - the characters exhibited subtle animations (the front character sat down to signal start, stood up to signal end of the 5-minute session). A ‘Chill Session’ calming nature scene was built into the application so participants could pause and decompress if needed during exposure.

Six adults who stutter were recruited from the Nottingham Self Help Group for People Who Stutter and completed Session 1; four of the six returned approximately two weeks later for Session 2. Each session involved a 5-minute prepared-topic speech to the virtual audience (‘Favorite Holiday’ for Session 1, ‘Job’ for Session 2). Outcome measures were:

• Self-reported anxiety on a 1-5 scale (1 = least anxious, 5 = most anxious), captured before, during (at 2:30), and after the exposure. This is NOT a validated social-anxiety instrument - it is a single-item Likert scale the author developed for the project.
• Body temperature and electrodermal activity (EDA) via an Affectiva Q Sensor worn on the wrist, captured at one-minute intervals during exposure plus baseline before and after.
• Qualitative speech observation by a ‘speech therapist/trained observer’ on the participant’s speech, face, mouth, breath, head, and body during the exposure.

In Session 2, the headset interface was updated with an Interface Scene allowing participants to adjust focus before entering the Lecture Hall Scene, and the lecture hall itself had its size, assets, lighting, and character animations adjusted relative to Session 1.

What they found

The paper’s Results section reports mixed outcomes across participants. Quoting the paper verbatim:

“The results were mixed, as for some participants, the VRET had no impact on their anxiety levels, for a few the VRET had an impact on their anxiety and their anxiety levels decreased as the sessions progressed and for others, the VRET had an impact on their anxiety and their anxiety levels increased as the sessions progressed.”

Only one participant’s anxiety table (P2) is shown in the paper, reporting Before=1, During=4, After=2 (Session 1) and Before=1, During=2, After=2 (Session 2) - a decrease in peak anxiety from Session 1 to Session 2 for that one participant.

Body temperature and EDA generally rose during exposure across sessions, consistent with physiological arousal in the speaking task. The observer noted that “as the sessions progressed into VRET Session 2, the participants showed signs of speaking more fluently and stuttering more fluently” - the wording ‘stuttering more fluently’ reflects a stuttering-modification (not stuttering-elimination) framing.

Participants reported that the Chill Session was useful. Participants also commented (after the sessions) that they felt more confident about the prospect of public speaking - though this was captured via the Post VRET Question Sheet open-ended responses, not a validated questionnaire.

Why this matters - and what to be careful about

This was a 2016 demonstration that consumer-grade mobile VR could be used to deliver a structured public-speaking exposure scenario for adults who stutter. As a conceptual precursor to subsequent immersive-VR-for-stuttering research (Brundage et al. 2006/2007/2015/2016, Moïse-Richard 2021, Chard et al. 2023, Delangle et al. 2026), it has historical value.

But several important caveats apply:

1. Mixed results. The paper does not show uniform anxiety reduction. Some participants showed decreased anxiety, some unchanged, some INCREASED. The current paper cannot be cited as evidence that VR exposure reduces anxiety in people who stutter.

2. No validated outcome measure. A 1-5 single-item anxiety scale is not a validated social-anxiety instrument. The standard measures used in subsequent VRET-for-SAD research - LSAS-SR, SPS, FNE, SIAS - were not used in this paper.

3. No control condition. Single-arm pilot with no comparison group.

4. ‘Chill Session’ safety-behavior concern. Chard & van Zalk’s 2022 scoping review of VRET designs for social anxiety in people who stutter EXPLICITLY EXCLUDED this paper from their primary corpus, citing both the lack of a validated social-anxiety outcome and the ‘chill session’ retreat option as a possible safety-behavior reinforcement. The inhibitory-learning framework (Craske et al. 2014) used in more recent VR-exposure protocols explicitly does NOT use retreat options, because they may interfere with the expectancy-violation learning that drives anxiety reduction.

5. Conference paper, not peer-reviewed journal. Bachelor’s dissertation work presented at iTAG 2016 (IEEE).

6. Sole-author / company-founder relationship. Gareth Walkom is both the sole author of this paper and the founder of withVR BV. The paper predates the company by years - it was Walkom’s bachelor’s dissertation project at Nottingham Trent University - but the self-citation pattern in an Evidence Hub published by withVR BV requires the most rigorous disclosure available, which this entry attempts to provide.

For Therapy withVR specifically: this 2016 paper did NOT use or evaluate Therapy withVR (which did not exist at the time). The system tested was a custom Samsung Gear VR application built by Walkom for the bachelor’s dissertation, with one environment, three avatars, a 1-5 anxiety scale, and the now-flagged ‘chill session’ retreat feature. The Walkom 2016 paper is included in the Evidence Hub as historical context for the development of VR-based approaches to stuttering anxiety, NOT as evidence of efficacy.

Limitations

• Sample size n=6 in Session 1, n=4 in Session 2. Very small; cannot generalize.
• No validated social-anxiety outcome measure. Only a 1-5 single-item self-report anxiety scale was used. Standard measures (LSAS-SR, SPS, FNE, SIAS) were not used.
• Mixed results across participants - decreased, unchanged, AND increased anxiety patterns were all observed.
• No control condition. Single-arm pilot.
• Single research site (Nottingham Trent University); single recruitment source (Nottingham Self Help Group for People Who Stutter).
• Conference paper (iTAG 2016 / IEEE), not peer-reviewed journal article.
• ‘Chill Session’ retreat option flagged by Chard & van Zalk (2022) as a possible safety-behavior reinforcement.
• No physiological-data fine-grained analysis beyond noting that body temperature and EDA generally rose during exposure.
• No long-term follow-up. Two sessions two weeks apart; no follow-up at 1 month, 6 months, or longer.
• No demographic detail beyond age range (sex, ethnicity, stuttering severity by validated instrument, prior treatment history, comorbid conditions not reported).
• Significant editorial COI. Sole author Gareth Walkom is the founder of withVR BV, the company hosting this Evidence Hub. The paper predates the company by years but the self-citation pattern requires the most rigorous disclosure available.
• Subsequent scoping-review exclusion. Chard & van Zalk’s 2022 scoping review (Front Digit Health, DOI 10.3389/fdgth.2022.842460) explicitly EXCLUDED this paper from their primary corpus on methodological grounds.