A Performance Evaluation of Nomon: A Flexible Interface for Noisy Single-Switch Users

Title of the Paper

A Performance Evaluation of Nomon: A Flexible Interface for Noisy Single-Switch Users

Paper Information

• Subject Area: Augmentative and Alternative Communication (AAC), Assistive Technology, Human-Computer Interaction
• Keywords: Assistive communication, accessibility, single-switch scanning systems, text input, noisy single-switch devices

Research Background and Problem Statement

• Problems and Challenges:

  1. Single-switch users (e.g., using button presses, puffing air, or blinking) experience low efficiency with traditional row-column scanning methods for text input, as they need to scan options row by row and column by column.
  2. Row-column scanning requires options to be arranged in a grid format, which is unsuitable for complex tasks such as gaming, drawing, and web browsing.
  3. Previous studies on single-switch input methods have focused on short time spans, lacking evaluations of long-term performance or applications beyond text input.
  4. Tests have traditionally been conducted using standard input devices (e.g., joystick buttons), without considering the reaction time characteristics of single-switch users with motor impairments.

• Significance: Single-switch devices provide a crucial means of communication for users with severe motor impairments, but their low efficiency and limitations significantly hinder user experience. Optimizing these devices in practical applications can enhance productivity and quality of life for this user group.

• Research Motivation and Related Work:

  1. Nomon is a flexible single-switch interface that uses probabilistic selection mechanisms to avoid the inefficiencies of sequentially cycling through options.
  2. Existing research suggests that Nomon may be faster and easier to use than row-column scanning, but lacks quantitative analysis over longer periods and broader application scenarios.
  3. Literature also discusses other methods used in conjunction with single-switch devices, such as the Dasher interface, but these methods have limitations.

Solution

• Methods and Innovations:

  1. Proposed an improved version of the Nomon interface, including interface adjustments to enhance accessibility, a calibration phase to optimize users' click timing distribution, and a simplified initialization process.
  2. Used simulation methods to optimize internal parameters of the Nomon interface, such as keyboard layout and the number of predictive completion options.
  3. Developed a webcam-based single-switch simulation technique to make the reaction times of non-disabled users more comparable to those of single-switch users with motor impairments.

• Implementation Steps and Key Technologies:

  1. Interface Optimization:

     • Made visual, interactive, and animation adjustments to the Nomon interface through collaboration with AAC experts and single-switch users.
     • Optimized the language model and keyboard layout through simulation experiments to improve text input efficiency.

  2. Experimental Design:

     • The experiment included two tasks: a text input task (spanning 10 sessions) and an image selection task (in the final session).
     • Compared the efficiency of Nomon and row-column scanning in different tasks.

  3. Real-Time Reaction Time Simulation:

     • Designed a webcam-based simulated switch requiring users to trigger the switch through head movements.
     • Verified in experiments that this simulated switch's reaction times were closer to those of actual single-switch users.

Research Findings

• Specific Results:

  1. Users achieved a 15% increase in words per minute when using Nomon for text input compared to row-column scanning.
  2. In the image selection task, Nomon users were 36% faster on average in making selections, though requiring more clicks.
  3. Nomon performed better in handling challenging text input (including out-of-vocabulary words), with significant improvements in error correction rates and input speed.

• Advantages of the Existing Solution:

  1. Nomon provides a more flexible user experience compared to row-column scanning, eliminating the need for a grid layout and expanding the range of applicable tasks.
  2. When dealing with rare words not covered by vocabulary prediction, Nomon demonstrated superior input accuracy and efficiency compared to row-column scanning.

• Experimental and Evaluation Results:

  1. Both non-disabled users and one motor-impaired user showed continuous improvement in performance with Nomon during the learning curve, while row-column scanning reached a performance plateau.
  2. Subjective feedback from users indicated that Nomon was easier to use and less error-prone, with 12 out of 13 participants ultimately preferring Nomon.

• Limitations and Future Directions:

  1. Currently, Nomon requires more clicks for larger option sets, necessitating further optimization, such as integrating eye-tracking technology.
  2. Plans to further validate Nomon's performance and usability, including testing with a broader range of motor-impaired users.
  3. Expanding the applicability of the Nomon interface to complex applications, such as smart home device control or dynamic web interactions.

Citation Information

• Source: CHI ’22, April 29-May 5, 2022, New Orleans, LA, USA
• DOI: 10.1145/3491102.3517738