- From: Will Pearson <will-pearson@tiscali.co.uk>
- Date: Mon, 29 Nov 2004 10:46:05 -0000
- To: "david poehlman" <david.poehlman@handsontechnologeyes.com>
- Cc: "Protocolls and formats" <w3c-wai-pf@w3.org>, <wai-xtech@w3.org>
Hi Dave; For a green square you could have something like: Line: 10, 10 20, 10 - color: green line: 10, 20, 20, 20 - color: green line: 10, 10, 10, 20 - color: green line: 20, 10, 20, 20 - color: green Admittedly, it's a rather large set of attributes to process in order to work out it's a green square. You would have to use something like the V Buffer to display them all, as according to George Miller's 1956 theory on short term memory, we can only remember between five and nine chunks of information in our short term memory. So, whilst it's accessible, as people can determine it's a gree square, it's not very usable, requiring an increased cognitive workload to determine the spatial relationships between the lines in order to determine the shape as being a square. However, if we look beyond speech as an output modality, it becomes a lot easier. As you're probably aware from my ASVS project, and to some extent, from the work of Evreinov and Meijer, you can display shapes using sound pixels. The basic concept is that you replace visual pixels of light with auditory pixels of sound. This just changes the communications channel used to convey the information, but will still allow the same perceptual techniques, such as the Gestalt laws of perception, to be applied to the auditory rendering as would be applied to the visual. So, you could determine that something was a line by having pixels grouped together with the same horizontal alignment for vertical lines, and the same vertical alignment for horizontal lines. Then by shothe lines in parallel, it becomes a lot easier and quicker to examine the spatial relationships between the lines and determine it's a square. Most attributes, such as font size, bold, italic, etc. are just differences in spatial relationship, and having this parallelism makes for easier determination of spatial relationships. As for color, well we usually have a hearing range from 20Hz to 20KHz. We're capable of detecting changes in tonal frequency at around the 10Hz to 15Hz mark, and so that would give us around 1300 to 2000 different states that could be signified through changes in frequency. Build in mechanisms for zooming to overcome the differences in auditory definition compared to visual definition, and something to simulate saccade movements, and you've got an auditory equivalent to visual output, well, what you actually have is the current thinking on my ASVS project *smile*, but it looks to work in theory *gsmile*. So, I think that if we look beyond speech, having the user extract the semantic meaning becomes a lot more usable. I noticed from Al's draft agenda for this week's PF, that tactile graphics are on there. Maybe this agenda item could be rescoped to include sonic graphics, although being output it's probably more in the domain of ua. Will ----- Original Message ----- From: "david poehlman" <david.poehlman@handsontechnologeyes.com> To: "Will Pearson" <will-pearson@tiscali.co.uk> Cc: "Protocolls and formats" <w3c-wai-pf@w3.org> Sent: Sunday, November 28, 2004 4:53 PM Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2 > Will, For most people, the leap is too hard to make. Can you send us an > example of what we'd have to extract meaning out of in text form? The > entire discussion to this point follows: > > Johnnie Apple Seed >
Received on Monday, 29 November 2004 10:45:14 UTC