First scoping review of immersive VR in speech-language pathology maps two decades of research

Background

Immersive VR has been adopted across many areas of rehabilitation - physical therapy, occupational therapy, mental health - but its place in speech-language pathology has been less clear. Individual studies exist across stuttering, voice, aphasia, dementia, and other populations, yet no review had systematically mapped how the field had been using immersive VR with people who have communication differences.

This scoping review by Nudelman, Niu, Hutz, and Edwards (all at the Department of Communication Sciences and Disorders, Syracuse University) is the first to do that mapping.

What the researchers did

The authors followed PRISMA-ScR guidelines for scoping reviews (Tricco et al., 2018; not pre-registered). They searched five databases (PubMed/MEDLINE, Science Direct, Web of Science, EBSCO, and Scopus) for peer-reviewed studies published between 1965 and 27 August 2025. The search string was developed using a machine-learning approach (litsearchr; Grames et al., 2019) - a documented AI-assisted method that builds keyword co-occurrence networks from an initial naive search to identify the most relevant terms.

Inclusion required three criteria: (a) immersive VR specifically (technology that creates a convincing sense of presence in a virtual environment - not screen-based or non-immersive 2D); (b) participants diagnosed with a communication disorder; and (c) involvement of a speech-language pathologist or CSD researcher in the study methodology. English-language peer-reviewed journal articles only.

Of 1,116 records identified from databases, 166 duplicates were removed, leaving 950 for title/abstract screening. The four authors independently screened all 950 records; Cohen’s kappa for interrater agreement at this stage was 0.341 (“fair”). 925 records were excluded; 25 were retrieved for full-paper review. Citation chasing of those 25 identified 11 additional records, bringing the full-text pool to 36. After full-text screening, 21 papers were excluded (17 for not using immersive VR, 4 for not enrolling patients with communication disorders, 2 for no SLP/CSD researcher involvement, and 4 across the citation-chase set), leaving 11 included studies.

The 11 studies, by population, were: Al-Nafjan et al. (2021), Brundage (2007), Brundage & Hancock (2015), Brundage et al. (2016), and Moïse-Richard et al. (2021) for stuttering; Franco et al. (2025) for aphasia; Hansa & Hansen (2025) for voice disorders; Harvey-Northrop et al. (2025) for communication disorders including language and cognitive-communication; Leyns et al. (2025) for gender-affirming voice training; Matsangidou et al. (2023) for dementia; and Vaezipour et al. (2022) for acquired neurogenic communication disorders.

What they found

The 11 included studies span 2007-2025, reflecting an upward trajectory in the field - eight of the 11 studies (73%) were published in the 2020s, mirroring the availability of consumer-grade headsets like the Meta Quest 2.

Population coverage. Stuttering dominated the literature: 5 of 11 studies (45%). The remaining six covered aphasia (Franco 2025), dementia (Matsangidou 2023), voice disorders (Hansa & Hansen 2025), gender-affirming voice training (Leyns 2025), acquired neurogenic communication disorders (Vaezipour 2022), and broader communication disorders including language and cognitive-communication impairments (Harvey-Northrop 2025). Participant ages spanned 9 to 81 years, reflecting the across-the-lifespan scope of speech-language pathology practice.

Methodology. Within-subject experimental designs were the most common (5 of 11, 45%). The remainder comprised qualitative interview studies (2), case studies (2), one pilot RCT (Leyns 2025), one single-group pre-post pilot (Matsangidou 2023), and one other design. Sample sizes ranged from 3 (Al-Nafjan) to 36 (Harvey-Northrop). Few studies incorporated randomization or blinding, and most lacked longitudinal follow-up.

Findings in stuttering. In stuttering, Brundage and colleagues showed strong correlations between stuttering frequency in VR and in matched real-world or clinical-interview settings. Brundage (2007) reported correlations of r = .904 (challenging VR job interview vs SSI-3 clinical interview) and r = .930 (supportive VR vs SSI-3). Brundage & Hancock (2015) reported %SS correlations of r = .82 (live vs neutral VA) and r = .99 (live vs challenging VA), plus PRCA-24 and PRCS correlations between live and VR conditions. Brundage et al. (2016) found that self-reported distress was higher in a virtual audience than a virtual empty room (medium effect size r = .57) while heart rate, skin conductance, and stuttering frequency did not differ. Moïse-Richard et al. (2021) showed strong SUDS correlations between virtual classroom and real audience speeches (ρ = .92, p < .001); virtual classroom and real audience SUDS did not significantly differ. Al-Nafjan et al. (2021) used automated stuttering event detection and qualitative debriefs with three participants and reported feelings of fear and immersion comparable to real-world speaking activities.

Findings beyond stuttering. Franco et al. (2025) compared VR-based and digital-static interventions for anomia in post-stroke aphasia. Overall picture-naming accuracy did not differ, but VR produced a significantly greater reduction in lexical errors (F(1, 65.58) = 20.02, p < .001). Matsangidou et al. (2023) reported high presence (M = 6.4/7), calming heart-rate responses (decreases from pre- to during-VR, p = .03), and significant reductions in negative affect during a 15-minute VR session in people with mild-to-severe dementia. Hansa & Hansen (2025) found that voice-therapy patients viewed a prototype VR system as therapeutically beneficial but emphasized that they “would not want to use the VR without therapeutic support.” Leyns et al. (2025) reported within-group fundamental-frequency increases for reading during VR voice training (d = 0.57) and a medium between-group effect favoring VRT (d = 0.76), alongside improvements in self-reported willingness-to-communicate. Harvey-Northrop et al. (2025) described an interprofessional collaboration to design a VR rehabilitation environment for SLP, focusing on process rather than patient outcomes. Vaezipour et al. (2022) reported average usability (SUS = 60.75/100), low NASA-TLX workload, and minimal simulator sickness in 10 adults with acquired neurogenic communication disorders, alongside semi-structured interviews identifying barriers including controller usability, headset comfort, and individual factors (claustrophobia, motor limitations).

Hardware evolution. The Meta Quest 2 was the most commonly used device (36% of studies), with older tethered HMDs (VFX-3D, eMagin z800) appearing in earlier work. The shift to wireless consumer-grade headsets has made immersive VR more practical for clinical research.

Application patterns. The authors note that across the included studies, immersive VR was most commonly used as a contextual manipulation to elicit ecologically valid speech and/or emotional responses (Al-Nafjan 2021, Brundage 2007/2015/2016, Moïse-Richard 2021). Relatively few studies used VR as an active, structured treatment platform with repeated practice, gaming, or explicit skill training (Franco 2025, Hansa & Hansen 2025, Leyns 2025). Biofeedback and adaptive task difficulty within VR were “largely absent” from the reviewed studies.

Why this matters

For the Evidence Hub specifically, this is the first review-level synthesis of immersive VR in speech-language pathology. Before this paper, clinicians and researchers asking “what does the VR-in-SLP literature actually look like?” had to piece it together from individual primary studies. Now there is a single peer-reviewed source that maps the field.

For practice, the review confirms that immersive VR can produce clinically meaningful approximations of real-world communicative responses. That ecological validity is exactly what is hard to achieve in traditional clinic-based therapy - and exactly what motor learning theory (Schmidt & Lee, 2011) says is needed for skills to transfer. The Hansa & Hansen (2025) finding that patients want VR delivered with - not in place of - clinician support is a reminder that VR is a tool for therapy, not a substitute for the therapist.

For research, the review identifies where the field needs to grow: bigger samples, longitudinal designs, randomized trials with blinding where possible, broader population coverage (especially preschool stuttering, language disorders, and cognitive-communication impairments), clearer operationalization of how VR is used as a therapeutic tool versus a contextual manipulation, and integration of biofeedback or adaptive difficulty - applications the authors describe as “largely absent” from the existing studies.

Limitations

The authors are transparent about the limits of their own work and of the literature it synthesizes:

• Interrater reliability during title/abstract screening was only fair (Cohen’s kappa = 0.341), suggesting subjective judgment may have influenced inclusion decisions. The authors suggest future reviews consider using AI screening tools as an adjunct.
• English-only inclusion may have excluded relevant non-English research.
• The included studies are mostly small, exploratory, and short-term; few employed randomization, blinding, or longitudinal follow-up.
• Heterogeneity in diagnostic groups, intervention protocols, and outcome measures makes cross-study synthesis difficult.
• Underrepresentation of language differences and cognitive-communication differences relative to stuttering.
• Inconsistent demographic reporting in some primary studies (e.g., participants’ gender not reported in Hansa & Hansen 2025 and Harvey-Northrop 2025; ages not reported in Harvey-Northrop 2025).

How this fits with the wider Evidence Hub

Eight of the 11 studies included in this scoping review are already summarized in the Evidence Hub as standalone entries (Al-Nafjan 2021, Brundage 2007, Brundage & Hancock 2015, Brundage 2016, Moïse-Richard 2021, Hansa & Hansen 2025, Leyns 2025) - linked via relatedStudySlugs above - plus the Brundage 2006 job-interview paper that is closely related to the cited Brundage 2007 tutorial. Four of the 11 - Franco 2025 (aphasia), Matsangidou 2023 (dementia), Harvey-Northrop 2025 (interprofessional-collaboration case study), and Vaezipour 2022 (acquired neurogenic communication disorders) - do not yet have standalone entries here. This scoping review now sits as the meta-level reference that ties the field together and contextualizes the existing entries.