The design and evaluation of non-visual information systems for blind users
This research was motivated by the sudden increase of hypermedia information (such as that found on CD-ROMs and on the World Wide Web), which was not initially accessible to blind people, although offered significant advantages over traditional braille and audiotape information. Existing non-visual information systems for blind people had very different designs and functionality, but none of them provided what was required according to user requirements studies: an easy-to-use non-visual interface to hypermedia material with a range of input devices for blind students. Furthermore, there was no single suitable design and evaluation methodology which could be used for the development of non-visual information systems. The aims of this research were therefore: (1) to develop a generic, iterative design and evaluation methodology consisting of a number of techniques suitable for formative evaluation of non-visual interfaces; (2) to explore non-visual interaction possibilities for a multimodal hypermedia browser for blind students based on user requirements; and (3) to apply the evaluation methodology to non-visual information systems at different stages of their development. The methodology developed and recommended consists of a range of complementary design and evaluation techniques, and successfully allowed the systematic development of prototype non-visual interfaces for blind users by identifying usability problems and developing solutions. Three prototype interfaces are described: the design and evaluation of two versions of a hypermedia browser; and an evaluation of a digital talking book. Recommendations made from the evaluations for an effective non-visual interface include the provision of a consistent multimodal interface, non-speech sounds for information and feedback, a range of simple and consistent commands for reading, navigation, orientation and output control, and support features. This research will inform developers of similar systems for blind users, and in addition, the methodology and design ideas are considered sufficiently generic, but also sufficiently detailed, that the findings could be applied successfully to the development of non-visual interfaces of any type.