[uf-discuss] Appeal for Issues: Empty spans in value-excerption-pattern

Isofarro lists at isofarro.uklinux.net
Sun Nov 9 09:32:49 PST 2008

André Luís wrote:
> I can see this option being adopted by people who are concerned about
> accessibility, while others might just go ahead and use the
> regular-problem-maker-of-a-pattern the datetime design pattern,
> preferring simplicity over accessibility. Right?

There are three possible configurations in a screen reader for handling 
title attributes on elements:

1.) Ignore the title attribute and read the inner text
2.) Read the title attribute instead of the inner text
3.) Read the title attribute as well as the inner text

The choice between these options is user choice - the user opts for the 
one that proves more useful over time.

In the situation where an empty span is used, with a class of value and 
machine data in the title attribute, this is read out where the screen 
reader is configured to use 2.) and 3.)

Where a value-classed span is used and a human friendly wording is used 
as inner text, then in case 2.) the machine formatted data is read out, 
but not the human-readable version, and in case 3.) the machine 
formatted data is read out before or after the human friendly data.

So the accessibility barriers that are created are:

1.) machine-formatted data is being read out to screen reader users
2.) machine-formatted data is being read out, and its human digestible 
format isn't.

Both cases result in content that is more difficult to understand, but 
case 2 is actually worse - it replaces human readable content with 
machine readable content. Both introduce accessibility barriers, just 
one does more damage than the other.

With this in mind, from an accessibility perspective, any microformat 
pattern which results in machine-formatted or human-unfriendly content 
in an area that is supposed to be human consumable is going to create 
one barrier or the other, depending on screen reader configuration. So 
the logical approach to protecting the accessibility of the page in 
these cases is not to use any microformat that specifies adding machine 
data into a human-visible region of the page.

More information about the microformats-discuss mailing list