A Usability Study of Nomon: A Flexible Interface for Single-Switch Users
Nicholas Bonaker, Emli-Mari Nel, Keith Vertanen, Tamara Broderick · 2023 · Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2023) · doi:10.1145/3597638.3608415
Summary
This paper presents the first usability study of Nomon — an alternative single-switch interface — with actual motor-impaired users, comparing it against the widely used row-column scanning (RCS) method. Single-switch interfaces are essential communication tools for people with severe motor impairments caused by conditions like cerebral palsy, muscular dystrophy, spinal cord injury, and stroke, where a switch might be activated by a blink, facial movement, thumb press, or puff of air. Unlike RCS, which requires options to be arranged in a rigid grid and uses sequential highlighting, Nomon places a small rotating clock indicator next to each selectable option anywhere on screen. Users select an option by clicking when their target clock hand passes noon, and Nomon uses Bayesian probabilistic inference to determine the intended selection from a series of clicks — meaning a single mistimed click does not necessarily cause an error. The researchers worked with two AAC charities (SpecialEffect and the Ace Centre) to recruit seven participants with diverse motor impairments and switch types, including Buddy Buttons, EMG switches, Jelly Bean switches, eye-gaze with blink detection, laser beam switches, and air mice with thumb switches. They redesigned Nomon to be fully accessible via a single switch and created a new graduated tutorial. Participants completed both picture-selection tasks (60 icons) and text-entry tasks across multiple sessions.
Key findings
In the picture-selection task, four of six participants selected pictures faster with Nomon, and two had similar rates between interfaces. In text entry, results varied more: one participant typed substantially faster with Nomon, two were faster with RCS, and three had similar rates. However, click loads were consistently higher with Nomon in both tasks — participants needed an average of 1.18 more clicks per selection in picture selection. This increased click load was the primary complaint and was particularly burdensome for participants prone to erroneous clicks, such as one participant with acid reflux that triggered involuntary switch activations. Dead-time between clicks was substantially longer with RCS (mean 6.02 seconds vs. 2.82 seconds for one participant), which may explain why six of seven participants perceived Nomon as faster even when their measured entry rates were similar or slower. Correction rates were generally lower with Nomon in text entry. Six of seven participants preferred Nomon overall, citing that it felt faster and had better predictive text. Participants with no prior scanning experience needed approximately nine practice sessions (about 90 minutes total) to become comfortable with Nomon. The researchers also released the first public dataset of real switch-user interactions with Nomon.
Relevance
This study addresses a critical gap in AAC research by testing an alternative to row-column scanning — the dominant single-switch method for decades — with actual motor-impaired users rather than simulated impairments. The findings suggest Nomon is a viable and often preferred alternative, particularly for tasks involving non-grid layouts like symbol-based AAC boards, file browsing, or game interfaces. The probabilistic selection mechanism that tolerates timing errors is a significant advantage for users with inconsistent motor control. For accessibility practitioners and AAC specialists, the detailed individual case studies highlight how dramatically switch users' abilities and preferences vary — underscoring the need for customizable, flexible interfaces rather than one-size-fits-all solutions. The identified challenge of click load points to a concrete area for improvement through multi-modal input combining eye-gaze and switch scanning. The open-source implementation and public dataset lower barriers for future research.
Tags: single switch · augmentative and alternative communication · motor impairment · text entry · scanning · assistive technology · input methods