Making electronic documents usable by readers with disabilities
Accessible electronic documents let readers find, understand and navigate information using different sensory and physical abilities. Meaningful links, structured headings, image descriptions, simple tables and text alternatives to speech make the document usable beyond its visual appearance.
Readers and accessibility
Electronic documents designed primarily for sighted users can create barriers: a chart may convey information only through colour, or conference links may be unidentifiable to a screen reader. Other readers may be unable to control a mouse, hear video sound or hold a long sentence in mind. Understanding disability as a consequence of disabling environments shifts attention towards changing those environments so people can participate more fully.
Definitions of disability differ between legal and administrative systems, and conditions as varied as paraplegia, blindness, autism, diabetes, HIV, arteriosclerosis, dyscalculia and schizophrenia have been included. Many definitions connect impairments of body or mind with adverse effects on social functioning. Accessibility guidance commonly addresses four groups:
- Cognitive or learning: difficulties with memory, attention, reading and numeracy.
- Visual: blindness, low vision and colour blindness.
- Auditory: deafness and hearing loss.
- Motor: conditions affecting movement, such as spinal cord injury.
Differences in definition and reliance on self-identification complicate estimates of how many people have disabilities. Sensory, motor and cognitive or learning impairments affect a substantial part of the population, particularly older people, making document accessibility important for reaching a broad audience and especially relevant to patient education.
Meaningful links
Underlined blue text provides a visual cue that cannot by itself identify a link for someone using a screen reader. Screen reader users may call up a list of links, so unlabelled links and generic labels such as “click here” or “link” make the desired destination difficult to find.
Link text should state its purpose, for example “Article on Links and Hypertext” or “Article on Contrast and Colour Accessibility”. Good labels are:
- Concise: brief but comprehensive.
- Descriptive: a summary of the link's specific purpose.
- Distinctive: unique rather than repeated for other links.
URLs should not serve as labels because they are neither brief nor descriptive and can be hard to understand through a screen reader.
Structured headings
Headings help readers scan topics and should be arranged in hierarchical levels. Manually changing size, colour or indentation may create the appearance of a heading without allowing a screen reader to recognise it or include it in a heading list.
Screen readers can recognise headings made with Microsoft Office's heading styles. Heading levels should not be skipped unless the style guide requires it; Heading 2 should not be passed over simply because Heading 3 looks better, since its appearance can be changed.
Alternative text and image purpose
Alternative text should make an image's information available to people who cannot see it, describing the content fully and accurately. A journal figure legend may already contain an appropriate description. Two questions guide the wording:
- What is the image's purpose?
- Given that purpose, how can its content best be described fully and accurately?
The surrounding text introducing the image provides context for those decisions. An example, introduced by a discussion of sex development, compares four descriptions of a karyotype showing a male with an extra X chromosome:
- “A karyotype of a male with an extra X chromosome”.
- “An image of an XXY male”.
- “A karyotype of a male with Klinefelter's syndrome (47, XXY)”.
- “A karyotype of a male with a disorder of sex development (an XXY male)”.
The first describes the image briefly and accurately without introducing a syndrome name or repeating nearby explanatory text. The second adds unnecessary “image of” wording; the third adds a diagnosis that non-experts cannot infer from the picture alone; and the fourth repeats the adjacent discussion of disorders of sex development.
Alternative text should usually be a short sentence or less, omitting unnecessary “image of” or “graphic of” wording and avoiding repetition of nearby text. Complex images, including graphs, maps and data-rich bar charts, may need another way to convey their information, such as a simple data table linked from beside the chart. A photograph can also be complex depending on its purpose; in 2022, NASA supplied lengthy descriptions for James Webb telescope images.
Simple data tables
Screen readers can read simple data tables without merged, split or empty cells. Multiple rows or columns of headers should be avoided because they can make tables difficult to read with a screen reader.
Captions and transcripts
Spoken material in conferences, podcasts and videos can present barriers to people with auditory or cognitive or learning impairments and should also be available as text. Audio can assist visually impaired podcast users, while text captions or transcripts assist those with hearing loss. The text alternatives have two forms:
- Captions: text showing what is being said, displayed on the screen.
- Transcript: the complete spoken narrative in a separate document.
Live captions can be produced through voice writing, also called respeaking, or communication access real-time translation (CART). A voice writer repeats blocks of speech with punctuation into recognition software, then corrects the output while listening to the next block. The process described in 2022 took about four to ten seconds per block, with the delay to corrected captions termed latency. More correction increases latency, and long delays can distract users. Delaying the presentation so the captioner hears its audio before the audience can reduce the perceived delay, while paraphrasing can shorten it at the cost of accuracy.
CART uses a stenotype machine and specialised software supported by an uploaded word database; the captioning accuracy target described in 2022 was at least 98%. Voice writers and CART captioners can also create transcripts. In 2022, Word 365 could transcribe microphone dictation and Office 365 Online could transcribe uploaded audio files; both required proofreading to correct errors.
Further guidance and training
Resources listed in 2022 included:
- Microsoft's “Make your Word documents accessible to people with disabilities”, covering headings, images, tables, links, lists, contrast and the Word Accessibility Checker.
- WebAIM Articles, covering accessible headings, links, lists and web pages.
- WebAIM's five-week Accessible Documents course for Word, PowerPoint, Excel and Acrobat PDF documents.
- The World Wide Web Consortium Web Accessibility Initiative's free Digital Accessibility Foundations course.