[Sigia-l] Accessibility and link design

[Thomas.Maynard at corp.terralycos.com]

Foks,

I've been doing some research into web accessibility for the visually
impaired, and have found something that is perhaps worth discussion.

Users of screen reader software (e.g., JAWS) have developed a few different
strategies for parsing a web page: they may
  1) use the brute force method of reading every element of the page, one
at a time, to develop a clear understanding of its structure and content,

  2) navigate across the links on the page using a key that will skip
intervening content

  3) access a dialog that lists all the links in a list box, giving them
the ability to directly access a link using the first letter in the text
string

Depending on the task, they may use any one of a number of strategies not
listed here. But if their task is to find a link that will lead them deeper
into the site and its content, they wil often focus exclusively on the
links. Their use of the links depends on the following:

1) the link must convey its function independently, since the user may not
have access to the surrounding content. This is problematic when the link
is a single word embedded in a block of text.

2) the link should provide a unique identifying element or keyword within
the first few words, preferrably the first. There appear to be a large
number of users who will only let the first part of the link be spoken
before moving on to the next link, or who use the list of links dialog and
try to access a link by guessing the first letter of the link.

These two facets of the link design seem to contradict each other. If a
link is designed to be clearly understood when read, the unique part of the
link may not be the first word of the link. For example, a link that says
"search news for cameras"  reads well, but if the user is thinking they
need to find a link for "news" or "cameras", they may not easily pick that
link out of a list of other links. However, a link that says "news search
for cameras" or "cameras news search" may have signifiacnt cognitive load
associated with them.

I hope I've been fairly clear in discribing this issue; if not please feel
free to ask for clarification

Tom Maynard
Sr. Information Architect
Terra Lycos