A common mode of providing access to computers has been to design screen-reading programs, which allow a user to read different parts of a computer display using text-to-speech. Such screen-reading programs are currently used to provide access to electronic information as well. Typically, the information is displayed on the visual display, and the screen-reading application is used to listen to the text. This approach does provide access to plain textual information. However, it breaks down when presenting technical information, e.g., written mathematics, tables, and other forms of highly structured information.
This break-down is a consequence of visually formatted information presenting the underlying structure using purely layout-oriented cues. When a screen-reading program speaks such visually laid out information, the underlying structure is not conveyed. To give an example, consider a table of numbers. When this is spoken by a screen-reading program (a line at a time), it is impossible to perceive the underlying structure.
In general, this kind of break-down happens constantly when structured information is spoken by a screen-reading program. When we read visual documents, we perform structured browsing. In general, this is impossible when using the traditional paradigm of displaying information visually and reading aloud this display.
Note that though the previous paragraphs are critical of the approach taken by traditional screen-reading programs, they should be regarded not as a criticism of the screen-reading paradigm itself. Screen-readers are designed to give access to computer applications and are appropriate1 for the tasks for which they were originally designed. What we point out here is that it is incorrect to overload the screen-reading programs with tasks for which they were not designed. There is a distinct shift in paradigms between providing access to a visual display and presenting structured information orally, and it is this shift in paradigm that we emphasize.
How structured information is presented orally should not be constrained by the way it is displayed visually. Visual information is laid out in a manner most suited to the visual mode of interaction; oral information should be presented in a manner most suited to oral communication.