The Migratory Cursor: Accurate Speech-Based Cursor Movement by Moving Multiple Ghost Cursors Using Non-Verbal Vocalizations
Yoshiyuki Mihara, Etsuya Shibayama, Shin Takahashi · 2005 · Proceedings of the 7th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '05) · doi:10.1145/1090785.1090801
Summary
This paper presents the migratory cursor, a novel voice-controlled cursor movement interface that combines two complementary techniques to achieve both speed and accuracy. The fundamental challenge with speech-based cursor control is that existing approaches are either fast but imprecise (discrete methods like grid-based systems) or accurate but slow (continuous methods where the cursor moves at a fixed rate until stopped). The migratory cursor solves this by combining both approaches. The system displays multiple numbered "ghost cursors" aligned horizontally or vertically with the actual cursor, spaced at regular intervals across the screen. The user first issues a direction command (e.g., "Move right"), causing ghost cursors to appear in that direction. The user then speaks the number of the ghost cursor nearest their target (e.g., "Seven") for a quick approximate positioning. Next, ghost cursors appear in the perpendicular direction, and the user repeats the process. For fine refinement, the user makes a non-verbal vocalization ("Ahhhhh...") which moves all ghost cursors continuously at a fixed pixel rate — the cursors stop when the user becomes silent. This eliminates the three types of delay that plague verbal stop commands: visual perception delay, speech production delay, and speech recognition delay. The system uses a volume threshold to distinguish non-verbal vocalizations from verbal commands.
Key findings
Three experiments with five able-bodied participants validated the approach. Experiment 1 (discrete specification) confirmed that distance between ghost cursors and target affected selection time as expected, but unit length (spacing between cursors at 16, 20, or 24 pixels) did not significantly affect time, and no subjects made errors. Experiment 2 (continuous specification) showed that non-verbal vocalization allowed significantly more accurate cursor positioning than verbal "Move"/"Stop" commands — subjects could stop the cursor within a few pixels of the target. Speed, distance, and command type all significantly affected accuracy, with faster cursor speed and non-verbal input producing shorter error distances. Crucially, non-verbal stopping was more precise because silence halts cursor movement immediately, eliminating the delays inherent in verbal stop commands. Experiment 3 (combined system) showed that all independent variables — unit length, distance, and target size — significantly affected total target selection time, but no subjects made errors when selecting targets. The distance from cursor to target did not greatly affect pointing time because the discrete specification handled large movements quickly, with continuous refinement only needed for the small remaining distance. All five subjects reported that while non-verbal vocalization was clear and effective, continuous vocalizing became tiring, especially with smaller unit lengths.
Relevance
This research addresses a critical accessibility need — enabling precise cursor control for people with physical disabilities who cannot use a keyboard or mouse. The migratory cursor's hybrid discrete-continuous approach represents a thoughtful solution to the speed-accuracy tradeoff that has limited speech-based cursor systems. The key innovation of using non-verbal vocalization for continuous movement, where silence immediately stops the cursor, elegantly sidesteps the recognition delays that make verbal stop commands imprecise. While the study was conducted with able-bodied participants (a limitation the authors acknowledge), the technique was explicitly designed for users with physical impairments. The finding that continuous vocalization becomes tiring points to an important usability constraint for real-world adoption. The work builds on and improves upon earlier grid-based and continuous cursor movement systems, and the authors discuss future directions including polar coordinate specification and zooming techniques that could further improve the approach.
Tags: cursor control · voice interface · speech technology · motor accessibility · alternative input · assistive technology · non-verbal interaction