Measuring the effectiveness of word prediction: The advantage of long-term use

This paper provides results regarding keystroke savings and text composition time from a long-term evaluation of a Swedish word prediction program. The evaluation was carried out as a single-case pilot study for one participant with dyslexia and no motor disabilities. During 13 months of training, during which a speech synthesis facility was introduced, his performance was shown to improve both with and without the word prediction program after an initial increase in text composition time. On average, he was able to save 22.7% total keystrokes with word prediction only and 30.5% with word prediction combined with speech synthesis. Text composition time increased upon introduction of the word prediction program by at least 22 %. Over time, however, and with the further introduction of speech synthesis, he was eventually able to return to his original text entry speed.