Dit4dah: Predictive Pruning for Morse Code Text Entry

Kumiko Tanaka-Ishii, Ian Frank

Abstract: We present a new predictive pruning algorithm for text entry and show empirically how it outperforms simple text prediction. Our tests are based on a new application domain for predictive entry: the input of Morse code. Our motiviation for this work was to contribute to the development of efficient entry systems for the seriously disabled, but we found that the constraint of using a single key highlighted features of text prediction not previously closely scrutinised. In particular, our tests show how predictive text entry is affected by two factors: altering the rankings of completion candidates based on the difficulty of entering the remaining text with just the keyboard, and the number of candidates presented to the user.