Identifying an Aurally Distinct Phrase Set for Text Entry Techniques

Abstract

In the last decade, interest in accessible and eyes-free text entry has continued to grow. However, little research has been done to explore the feasibility of using audibly distinct phrases for text entry tasks. To better understand whether preexisting phrases used in text entry research are sufficiently distinct for eyes-free text entry tasks, we used Microsoft’s and Apple’s desktop text-to-speech systems to generate all 500 phrases from MacKenzie and Soukoreff’s set [32] using the default male and female voices. We then asked 392 participants recruited through Amazon’s Mechanical Turk to transcribe the generated audio clips. We report participant transcription errors and present the 96 phrases that were observed with no comprehension errors. These phrases were further tested with 80 participants who identified as low-vision and/or blind recruited through Twitter. We contribute the 92 phrases that were observed to maintain no comprehension errors across both experiments.

Award
Honorable Mention
Authors
Jacob Abbott
Indiana University Bloomington, Bloomington, Indiana, United States
Jofish Kaye
Mozilla, Mountain View, California, United States
James Clawson
Indiana University Bloomington, Bloomington, Indiana, United States
Paper URL

https://dl.acm.org/doi/abs/10.1145/3491102.3501897

Video

Conference: CHI 2022

The ACM CHI Conference on Human Factors in Computing Systems (https://chi2022.acm.org/)

Session: UX Methodology

288-289
5 items in this session
2022-05-03 09:00:00
2022-05-03 10:15:00