Rebuilding The Web

Articles, advocacy, discussion and debate about the many problems of the Web and the challenges of rebuilding it.

Do we need three ways to describe images?

Many people interpret the HTML spec to define alt as a short description of an image and longdesc as a detailed description. Is this useful? This definition of alt does not take into consideration the context of the image, making comprehension difficult when reading content containing images that cannot be seen. The definition of longdesc is also problematic, because content authors see little point in writing two versions of a description whose only difference is their length. So, what is it that we really need?

Yes, we need three ways to describe images

First, we need a replacement for images when images cannot be seen. Content in the alt attribute fulfills this role nicely. Alternate text is displayed in place of an image with the objective of ensuring that it works within surrounding content. The flow of the document is uninterrupted and cohesive, and comprehensible when images cannot bee seen.

Second, we need a description for images. For those that cannot see images or can only see some aspects of images (due to color blindness for example), image descriptions offer the only means of understanding what is in an image. Even with its shortcomings, longdesc fulfills this role. But descriptions have advantages even when images can be seen, because descriptions allow images to be described in different formats (such as tabular data of a graph), and they can supply information not readily perceived by glancing at an image, such as the name of a person in a photograph.

Third, some Web users need textual glances of images, because sighted people tend to glance at webpages in order to determine if they should commit to reading the content, and users with limited or no vision would like to be able to do the same. So they need short textual descriptions of images. Alternate text does not fit this role because it is affected by context (by surrounding text), which is why the same image can have different alternate text in different parts of the document. Textual glances of images however should not be affected by context.

Do we need a new HTML feature for textual glances?

Maybe. The data to be used for textual glances is already part of HTML. It's actually the title of the document referenced by the longdesc attribute. This could serve the need for a textual glance when longdesc points to an external document - a document dedicated to describing an image. But this feature fails if longdesc points to content within the same document.

Conclusion

We need three types of image descriptions: a textual replacement, a textual description, and a textual glance. Alternate text does not work for textual glances because alternate text changes with its context, whereas descriptions and textual glances of images are not affected by context. Textual glances could be extracted from the title of external documents referenced by longdesc.

Public comments

1. Posted by mattur
on Tuesday 2011-05-10 at 02:30:25 PST

No

Comments are closed for this article.

Main menu

Check out the a11y bugs project that aims to help browser / tool vendors fix accessibility bugs.