Microsoft Research have published a really interesting piece of user research on Alt Text image descriptions. The paper is “Person, Shoes, Tree. Is the Person Naked?” What People with Vision Impairments Want in Image Descriptions.
Describing what images represent matters for people using screen reading software. Otherwise the meaning of a web page or service gets lost in the gaps between text and visual elements.
The new research is interesting because it finally asks people with a range of visual impairments what they want from image descriptions.
This is about adding context to alt text.
The above table is from the research and usefully frames how to think about user centred image descriptions.
The user intent is primary. What is the person doing? Shopping, working, dating, reading the news or checking a social network perhaps?
The second part is what is to be described. Three main themes:
In what ways these are described differs.
I have tried to visualise the research in a non-table format so as to understand differences. Here are four images.
Very little description of Scene but lots and lots about the Object for sale.
The Scene and People are priorities.
Unsurprisingly, it’s about the personal description in detail. However, some Objects and the Scene may be needed to add context.
It appears that everything might be relevant. Social Networking covers a lot of possible subjects and contexts.
What might work better?
This is all just a quick attempt on visualising information from the Microsoft research.
‘I am not a designer’ and it would be interesting to see better ways of making the information better known.
This research is important as it is based on identified user needs.
Adding context to alt text matters to accessibility.