LOVIE: a word prediction list optimized for visually impaired users on smartphones

Abstract

On mobile devices, most text entry systems offer assistance through the means of prediction or correction word lists. These lists can hardly be used by visually impaired users in a short time. In this paper, we present three studies to find an interaction which minimizes the time required to select a word from a list without knowing its elements a priori. First, we explored the various possible item layouts, to conclude that a linear spatial arrangement is preferable. Then we studied the audio feedback, and we have determined that providing on-demand audio feedback is enough. Finally, we proposed four validation interactions to retain a validation based on the principle of a press-release gesture.