Currently, most dictation strategies require the user to constantly monitor
the response of the dictation software, to correct mistakes and to improve
the system's performance.

This monitoring is usually done visually.  If this were to be done with a
screen reader, the process would be something like:
        1. user utters phrase
        2. computer makes guess and speaks guess out loud
        3. if guess is correct continue to step 1 with next phrase
        4. if guess is incorrect
                user requests computer to speak list of choices,
                then speaks a command to make a choice
                -or- user repeats utterance at step 1

Even after you tackle the problem of getting the screen reader to read
every prompt correctly, this speaking and then listening to choices for
every utterance gets pretty tedious.

For most blind users, the problem with computer interaction is not the
input of data or commands.  That is readily achieved through typing and the
memorization of keyboard commands.  The challenge is really in determining
the current state of the computer and what all of the user's options are in
the current state.

Ever since the advent of GUI's, the number of options available at any
given point in a computer program has grown enormously -- oweing to the
fact that a sighted person can quickly interpret well-designed visual
representations of the current state of the program and what the user's
options are at that point.

On the other hand, blind users still must operate a computer through wrote
memorization of what commands are available at any particular state of a
program -- similar to the way sighted people used command interfaces 15
years ago.

I'm not convinced there is any way to translate the well-designed,
two-dimensional, high-resolution representation of a program's current
state and options into one-dimensional speech or low-resolution tactile
graphics used by blind people.

It seems for now that blind users must rely on wrote memorization of
increasingly complex software -- certainly a disadvantage to the way
sighted users interact with software.

At 07:35 PM 7/1/98 -0400, you wrote:
>Hi, Folks:
>
>This sounds interesting, but I have no clue if it is doable.
>
>Reactions?
>
>Peter Altschul
>
>>X-Authentication-Warning: telepath.com: majordom set sender to
>[log in to unmask] using -f
>>Date: Wed, 01 Jul 1998 13:21:10 -0500
>>From: Sharon Geddes <[log in to unmask]>
>>To: [log in to unmask]
>>Subject: Nolan Crabb
>>Content-Disposition: inline
>>Sender: [log in to unmask]
>>Reply-To: [log in to unmask]
>>
>>+== acb-l Message from Sharon Geddes <[log in to unmask]> ==+
>>After watching me try, with difficulty to pick a screen reader, then
>>navigate about both Windows 3.11 and Windows 95, my assistant made
>>a suggestion.  Many efforts have been put into trying to get the
>>companies to make their products accessible.  Why not combine Dragon
>>Dictate with a screen reader into one program.  We could give the
>>computer commands vergbally, thus eliminating set files, macros, and
>>other difficulties.  We could also have a screen reader which would read
>>the computer screen to us.  This is just a suggestion, but I feel it is a
>>good one, and would like to see someone try it.
>>************************************************************
>>* ACB-L is maintained and brought to you as a service      *
>>* of the American Council of the Blind.                    *
>>************************************************************
>>
>>
>
>
Curt Griesel
U of Minnesota Academic & Distributed Computing Services
mailto:[log in to unmask]