"We need a vision first": Speculating Deaf-Centered Immersive Classrooms
Shuxu Huffman, Abraham Glasser, Christopher Hayes, Christian Vogler, Raja Kushalnagar · 2026 · Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA '26) · doi:10.1145/3772363.3798360
Summary
Huffman and colleagues at Gallaudet University present an all-Deaf-authored speculative design study that asks not how extended-reality (XR) classrooms can be retrofitted for Deaf students, but what XR classrooms should look like when Deaf teachers lead the design from scratch. The work is framed explicitly as a prequel to prototype-driven research: rather than testing a system against Deaf users, the authors ran a Deaf-led Future Workshop (a participatory method due to Jungk and Mullert that supports collective imagining and idea generation) with four Deaf teachers (DT1-DT4), three of whom are co-authors. Across two one-hour sessions conducted entirely in ASL and video-recorded, teachers moved through warm-up, individual envisioning, group co-design, and collective reflection phases. The first author then translated transcripts and sketches into written English and led reflexive thematic analysis, with all co-authors refining interpretations. The paper frames Deaf teachers as epistemic leaders rather than informants. Central commitments include treating Deaf pedagogy (shared visual attention, embodied signing, collective awareness, horseshoe seating, DeafSpace principles) as foundational rather than as accommodations layered onto hearing-centred instruction, and treating XR's value as the capacity to organise visual information in ways that are difficult or impossible in physical rooms, not as a way to replicate physical classrooms.
Key findings
Three interpretive design orientations emerged. First, DeafSpace in immersive environments is an ongoing coordination of shared visual access, not a fixed spatial layout. Deaf teachers envisioned XR classrooms where sightlines, signer positioning, and content placement dynamically reconfigure as discussion moves; individual customisation must not undermine shared understanding, and systems should help maintain alignment across viewpoints (for example flagging when a signer change or object move disrupts visual access for others). Second, sign communication in XR must be supported as embodied, relational interaction: full-body visibility, facial expression, spatial alignment with materials (pointing at slides while signing), and trusted representation of the signer. Teachers expressed strong preference for 'real person' representations over avatars, with the concern that avatars may flatten signing styles and undermine trust ('I prefer a real person, not an avatar' — DT4). Separating signer from content (signer on one screen, slides on another) was described as disruptive, since pointing, gaze, and body orientation are integrated with instructional material. Third, XR's most valuable contribution is visual orchestration — the deliberate coordination of visual cues, salience, and timing to support shared attention — rather than in-person replication. DT2 compared this to volume control, wanting features that 'turn up' the active signer or relevant content while 'turning down' visual clutter. DT4 put the orientation directly: 'We should not try to use it to replicate in-person interaction. Instead, we should focus on what VR does well.' DT1 anchored the thread: 'in the end, it is still about human-to-human connection. Technology should support that, not replace it.'
Relevance
For accessibility practitioners and educational-technology teams, this paper is a methodological exemplar and a substantive warning. Methodologically, it demonstrates what Deaf Tech research looks like when the research team, workshop participants, and interpretive authority are all Deaf: conventional participatory design that treats Deaf people as late-stage feedback providers produces technologies that fail to reflect Deaf epistemologies. Substantively, the orientations translate into concrete design requirements for any immersive classroom, video-call tool, or hybrid-meeting platform serving signing users — dynamic sightline management, synchronised public-to-personal visual spaces, signer-content integration rather than separation, signaling of visual disruption, and visual-salience controls analogous to audio mixing. The avatar concern is particularly timely as generative sign-language avatars proliferate: these Deaf teachers did not endorse them. Limitations are openly acknowledged: N=4, all from one Deaf educational institution; the work is early-stage and does not generalise; and the orientations are conceptual framings for future research rather than empirically validated design requirements. The paper is explicitly a prequel — conceptual grounding for prototype work that builds with Deaf communities from the outset rather than onto them.
Tags: Deaf Tech · extended reality · Deaf culture · Deaf education · speculative design · participatory design · DeafSpace · sign language · American Sign Language · accessibility theory