The project aimed to develop a set of
guidelines for ALT texts (useful to both sighted and blind people) that could be
adopted by web designers. Workable guidelines are necessary as ALT texts are not
standardised, nor are they an obligatory inclusion. The ALT texts should not
rely on ASCII art, as this is often meaningless when spoken by a screen reader.
Procedure.
After background research (looking at access technology,
existing ALT guidelines, and verbal-mental representations) a set of preliminary
guidelines were developed. These recommended strong syntactic structures, and
the possibility of using a non-speech sound ALT text interface. However, when
these guidelines were tested on an image laden 'mock' website, their ethos
changed altogether. Rather than using the ALT text for containing a concrete set
of information, they should be adaptable according to user preference.
Results and Recommendations.
The following recommendations were
formulated after online and offline testing:
Bullet Points: ALT ="*".
Logos: ALT="Logo company name", where the appropriate company name is
inserted.
Navigation icons: ALT=" destination", where an appropriate brief
destination is given.
Image links: ALT="description~ destination" where description is a brief
alternative to the image.
D links: D text links should read "Description of....". Longdesc tags were
deemed unreliable mechanisms.
Animated GIFs: there should only be one ALT, and this should convey an
appropriate sense of movement belonging to the image.
Examples of each type of image and their ALT are given below.
Description of Megalithic
Stone Circle.
Conclusions and Future Work.
The principal conclusion drawn from the
project is that each user wishes to have a different amount of information in
ALT texts. This could be achieved by:
Additional attributes of the IMG tag in HTML. These could contain
informations such as the size of the image, its importance etc. The user would
choose whether to display these.
Reconfigurable screen readers. Users could choose whether they wished the
screen reader to read information about image size and importance, or announce
~.
Future work.
Guidelines should be developed for imagemap ALT texts.
Investigations could be made into the combination of non-speech sounds and
ALT texts. At present there are many difficulties with streaming non-speech
sounds and the screen reader.
Investigation into memory and recall when browsing. Does the recall of
spoken word change significantly with different browsing patterns? If so,
should ALT texts be ordered in a specific way?