UC Merced

Text entry is one of the most commonly performed tasks on mobile devices, however, it still poses accessibility challenges. This dissertation investigates existing text entry techniques for tablet computers, smartphones, and smartwatches to identify accessibility issues with them, then designs, develops, and evaluates more inclusive text entry techniques to mitigate these problems. The first part of the dissertation presents Senorita — a novel two-thumb virtual chorded keyboard for sighted, low vision, and blind mobile users. The technique is design following the concept of inclusive design, using mainstream technology without any extramural devices. Results show that with Senorita blind users surpass their Qwerty entry speed by 32%. Besides, sighted users yield a comparative performance both on a smartphone and tablet. The second part of the dissertation explores a crown-based keyboard — a technique to enable text entry with one finger using the crown of a smartwatch for people with limited dexterity. It uses an iterative design process with multiple user studies. Results show that manual rotation is faster (4 vs. 6 wpm) and more accurate than automated for able-bodied people. A final study with ten people with limited dexterity reveal that both automated (2.3 wpm) and manual (2.6 wpm) the crown-based keyboards were faster and more accurate than the default Qwerty method on smartwatches. The third part of the dissertation presents SwipeRing — a novel keyboard that arranges the Qwerty layout around the bezel of a smartwatch divided into zones to enable gesture typing. These zones are optimized for usability and to maintain similarities between the gestures drawn on a mobile virtual Qwerty and a smartwatch to facilitate skill transfer. The comparison of SwipeRing with a similar method shows that SwipeRing yields a 33% faster entry speed (16.67 wpm) and a 56% lower error rate. Finally, the fourth part of the dissertation presents new action-level performance metrics: UnitER, UA, and UnitCX. They account for the error rate, accuracy, and complexity of multi-step chorded and constructive text entry methods. Results of a longitudinal study with two existing multi-step methods confirm that these metrics help to identify probable causes of slower text entry and input errors.