Reclaiming VR Design Authority: Deaf Signers Shaping Immersive Classrooms
Shuxu Huffman, Laura South, Matthew James Buckman, Raja Kushalnagar, Francisco Raul Ortega, Abraham Glasser · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3790499
Summary
This paper reports a Deaf-led, ASL-first study of a VR classroom prototype designed to reduce the persistent "visual attention split" that Deaf students experience in remote and online classrooms, where gaze must constantly shuttle between a signing instructor and instructional materials such as slides or code. The authors, three Deaf and three hearing researchers with leadership from Deaf scholars, explicitly frame the project as an exercise in reclaiming VR design authority: rather than adding a caption overlay or interpreter box to a hearing-normed classroom (what the paper calls an "after-the-fact retrofit"), they situate their prototype within Deaf epistemologies and Deaf pedagogy, treating visual primacy, spatial layout, direct signed instruction, and cultural resonance as first-class design goals. The Unity-based prototype runs on Meta Quest 3 and presents a 15-minute ASL lecture on binary search, taught by a Deaf instructor with slides and live-coding panels over a plain blue background. It implements three signer-placement modes users can switch between in real time: Corner Mode (signer in one of four screen corners, repositionable with a key), Parallel Mode (signer beside the content on the blue background), and Transparent Mode (signer overlaid semi-transparently on the content with adjustable opacity). Twelve Deaf participants (students, teachers, and staff, all ASL-fluent) explored all three modes and then gave semi-structured interviews in ASL. Analysis used Constructivist Grounded Theory with five stages - translation, initial coding, axial coding, selective coding, and theoretical integration - yielding 111 codes collaboratively validated by the Deaf co-authors.
Key findings
Participants reported that VR substantially reduced visual attention split compared with 2D video platforms such as Zoom, improved sign visibility and field of view, lowered the need to pause and rewind, and supported greater physical and postural flexibility than a fixed laptop setup. No single placement mode was universally preferred; each served different purposes. Corner Mode was the most broadly accepted, valued for flexibility and familiarity (it resembles Disney's ASL overlays); participants frequently repositioned the signer to avoid occluding live code or slides. Parallel Mode was more polarizing: some found the larger signer image and clear separation easier to read, while others found the increased distance between signer and content cognitively taxing. Transparent Mode was the most controversial - praised by some for collapsing signer-content distance, disliked by others due to insufficient contrast between the signer's skin tone and the background content, and for being visually unfamiliar. Some participants with cochlear implants found the headset physically uncomfortable. From these findings the authors synthesize a five-dimension conceptual framework for evaluating signer placement: proximity (visual distance between signer and content), customizability (ability to adjust signer size, position, opacity, or distance), visual efficiency (how easily sign and content can be processed together), cultural fit (alignment with Deaf pedagogical norms), and task flexibility (suitability across teaching activities). The authors map each mode's trade-offs across these dimensions.
Relevance
For accessibility practitioners this paper is significant for three reasons. First, it operationalizes Deaf Tech and Deaf-centered design methodologically, showing how community-led research can move beyond after-the-fact captioning and interpreter insertion. Second, the five-dimension framework is immediately transferable: anyone designing videoconferencing, online learning, live-streaming, or mixed-reality tools that include signed content can use proximity, customizability, visual efficiency, cultural fit, and task flexibility as evaluative lenses, and the framework's claim that "visual layout is a cultural and embodied practice" reframes what have often been treated as mere interface preferences. Third, the study directly challenges the dominant pattern in which slides and spoken content are treated as primary and the signer as peripheral; in Deaf classrooms, the signer is the primary channel, and interface design must reflect that. Limitations include the small (N=12) U.S. ASL-only sample, a single 15-minute lecture on one technical topic, no quantitative performance data (by design, to avoid hearing-normed testing pressure), current VR hardware discomfort (especially for users of cochlear implants and hearing aids), and keyboard-based interaction rather than hand-tracking. Future work should explore sustained deployments, other sign languages, diverse subject matter, and gaze- or gesture-based signer repositioning.
Tags: deaf and hard of hearing · deaf tech · virtual reality · visual attention split · american sign language · participatory design · education · deaf-centered design