Recognition accuracy and usability of pen-based interfaces

Questions whether devices capable of handwriting recognition are likely to appeal to those who have relatively little experience of computing, and have poor keyboard skills. The answer seems to be that pen computing can be an attractive prospect for this group of potential users, although it is the inability to type rather than lack of computer experience which matters. User attitudes appear to be driven by an assessment of the relative costs of using the device. This takes account not only of the effort required by non-typists to use the keyboard, but also of the perceived difficulty of writing legibly. As the experience of using the pen for text input alerted subjects to the consequences of recognition failures, this became a significant factor in the assessment of costs and benefits. At the same time, the focus of the evaluation also shifted from the input medium to the task. The inconvenience of writing in a particular style, and of correcting recognition errors, could be offset against the perceived benefits of using the pen within a particular application. The effect of this broader assessment of both costs and benefits was to reduce the significance of user characteristics. The highly positive evaluations of our application demonstrates that it is possible to design successful software for pen applications, despite the limited performance of current recognisers. Although new groups of users may initially be attracted by the idea of pen computing these applications are likely to be judged on their merits by all users, regardless of their existing skills