> The interaction would best take place at the level of the access software > so that the braille and speech devices can be configured to perform > complementary functions, in ways that would require exploration and > experimentation. I encapsulated the concept of multiple rendering paths of the same data in the term "polymedia" some years ago. The notion is that content (defined for presentation in structural tags) is parsed by presentation software and rendered in as many forms as available output devices allow, or any subset of that capability that the user has chosen to use. (Artistic control can be accommodated, but the topic is beyond what we need to consider here.) Polymedia differs from multimedia in that the user can reasonably expect to have access to data in all the forms that are most useful, rather than in a single controlled access path determined by the information publisher. And a user might use different forms at different times depending on the output devices available or the tasks that the user is performing. (Access while driving a car might very well require a different output medium from access while sitting at a desk, for example.)