Counting Fingers: Eyes-Free Text Entry without Touch Location

Keith Vertanen

CHI '16: Extended Abstracts of the ACM International Conference on Human Factors in Computing Systems (workshop), 2016.

Entering text on a touchscreen device without visual feedback can be challenging. In this paper I explore the feasibility of an approach where a single character of output is ambiguously specified by touching the screen with between one to five fingers. The count of fingers is fed to a sentence-based decoder which attempts to infer a user's intended text. This approach allows users to touch the screen with any finger orientation so long as it is touched with the correct number of fingers. I compare the accuracy of five different finger count to letter mapping, including two based on braille. Four of the mappings provided promising accuracy with character error rates below 5% on simulated input.