Imagine that a newly built public building doesn’t have step free access. Instead, a flight of stairs leads to the front door. The building exists for everyone to use but can’t be accessed by everyone because the stairs are a barrier. People using wheelchairs, other mobility devices, or pushchairs are all excluded. Digital content accessibility can be thought of in the same way. Just because a document has been published online doesn’t mean it can be used by everyone.
There is no doubt that creating accessible digital content is a good thing. Even if U.K. law didn’t now require public sector organisations to maximise their digital content accessibility it would still be the right thing to do. However, whilst we are passionate and proactive about embedding accessible and inclusive practices it hasn’t always been easy.
Here we explore the ups and downs of transforming documents from our institutional repository to maximise their accessibility, so they can be used by everyone.
How we got here
Our institutional repository (known as KAR) is where University of Kent researchers archive and share their publications and other research works. In September 2020 we added a clickable button to all open files in KAR so anyone can request an accessible version of the work.
We decided to adopt this reactive approach to accessibility because we don’t have the resources to proactively improve the digital accessibility of all 56,000 (and growing) records in our repository. Find out more about how we developed this accessibility process.
Overview of an accessibility request
We now receive accessibility requests every week. First, we confirm that the request is definitely needed. Accessibility can mean different things to different people so, by explaining that it is reformatted so that it can be used with assistive technologies, we ensure we are meeting the users need.
Once the accessibility request is confirmed we start by uploading a Word format of the document to Blackboard Ally. This software/application helps us to gauge the current accessibly of the document by giving a percentage measure and suggesting how we can raise that score. So far most documents have scored between 20% and 43% accessible which is very low.
We start to improve the overall accessibility of the document which can involve adding or editing:
- Navigable headings and subheadings.
- Alternative text descriptions to images, tables, graphs and diagrams so that they can be meaningfully interpreted.
- Meaningful hyperlinks that describe where the reader will go when they click on a link to an external website.
- Transcripts and captions for audio visual content such a video recording or blog.
- Contrast that increases definition between the background and foreground.
- Increasing font size to at least 14 and in ‘selectable text’ so that it can be copied and pasted.
- Reading order on tables so the header row is clearly defined.
How long does it take?
The time it takes to reformat a digital document to maximise its accessibility varies hugely depending on the type of document. So far a single document has taken anything from just five minutes and up to three weeks. The amount of time it takes varies so much because we are working with multiple document types that have different contents, are formatted with contrasting structures, vary in length, and range in their level of subject complexity, and which are originally created by authors and publishers.
Five minutes is the quickest we’ve been able to fully reformat a document. It was a nine-page, chiefly text-based article, with a structure that was straightforward to follow. We improved the document by:
- Increasing the font size
- Adding navigable headings
- Adding meaningful link text
- Setting the reading order of the table and adding alternative text.
The accessibility score increased from 48% to 100%.
When our first thesis accessibility request arrived recently, we knew it was going to take a little longer but we had no idea just how complicated it would be.
Thesis are produced by Master and Doctoral students at the end of their studies and can be large documents that contain text, diagrams, images, tables, scientific formulas, and other complex content.
The first thesis accessibility request we received was for a 186-page document that was just 47% accessible. Testing revealed some of the key accessibility issues:
- Headings not following a logical structure
- Images missing alternative descriptions
- Tables missing header rows and alternative descriptions
- Text with insufficient contrast
This information, however, did not reveal every accessibility problem.
Why did it take us three weeks to fix four issues?
- Because document testing doesn’t tell us everything. Having the ability to test the accessibility of a document is brilliant. We can measure inaccessibility and evidence the improvements we’ve made. However, if we don’t account for this shortfall we risk providing a document that remains inaccessible despite testing results being 100%.
- Because some issues appeared on every page and had to be corrected individually. For example, authors do not add headings, tables or images in a consistent manner and often pull their theses together from several separate documents created over the time of his research project.
- Because some of the required changes need specialist subject knowledge or qualitative choices that should be made by the author. In these cases, we try to contact the author for help. In this case the author had left the University and was living in America, so time-zone differences added to the communication issues. Sometimes we cannot contact the author at all.
Here are some of the issues that took a long time to address but were not identified by the accessibility testing software:
- Navigable headings
- Testing told us that headings used throughout the thesis did not follow a logical structure, e.g. they had in-built Style Headings such as ‘Title’, ‘Subtitle’, ‘Heading 1’, ‘Heading 2’, and so on, but were not used in a logical order.
- Testing did not recognise that some headings did not have any in-built ‘Style Headings’ applied to them. Instead, the headings were highlighted with just bold text and underlined.
- This created an accessibility barrier because users of read-aloud assistive technology will not be informed that a new heading, and therefore new section, is being read aloud to them.
- We fixed this by reformatting all headings into a navigable order and added Style Headings to those where none were originally applied.
- Tables saved as images
- Testing told us that all images were missing alterative text descriptions.
- Testing did not recognise that tables had been saved as images.
- This created an accessibility barrier because tables need a reading order applied so that users of assistive technologies know the order the tables need to be used. Tables saved as images cannot have a reading order applied because the formatting does not allow this. This makes the table meaningless to the reader.
- We fixed this by recreating the tables in the correct format, using the images of the original tables as guides, and applying a reading order to them. We also added captions and alternative text descriptions to the tables.
- Images, tables and graphs missing alternative text descriptions
- Testing told us that none of the images, tables, or graphs had alternative text descriptions.
- Testing did not recognise that, as we have already discussed, the formatting of some of the images was incorrect.
- This created an accessibility barrier because images and tables without alternative text descriptions are meaningless to those who rely on assistive technologies.
- We fixed this by adding alternative text descriptions to all of the images and tables. This was challenging because many of these images, tables and graphs included subject specific content used within specific contexts. We used text from the thesis to help us add accurate alternative text descriptions.
- Erroneous page footers and missing page numbers
- Testing told us that text boxes added throughout the document were not wrapped in line with the main text of the document. Testing also told us that the document did not have any page numbers.
- Testing did not recognise that multiple text boxes had been added to the page footers. This included page numbers but, because they had not been properly formatted, they could not be recognised through testing.
- This created an accessibility barrier because each text box in the page footer, including the page numbers, would not be properly interpreted by assistive technologies.
- We fixed this by manually removing the text boxes from the footers. This involved removing 558 text boxes as the automated procedure could not carry out this function. We chose to remove them because, other than the page numbers, the remaining text boxes in the footer contained duplicate text that was included elsewhere in the document and was not core. We then added formal page numbers using the functionality in Word.
- Table of contents entered manually
- Testing told us that heading orders were incorrectly formatted.
- Testing did not recognise that the table of contents used in the document had been created manually rather than using the automatic functionality in Word.
- This created an accessibility barrier because the document could not be easily navigated by anyone but especially those using assistive technologies.
- We fixed this by first fixing all of the erroneous headings throughout the document. We deleted the original table of contents and added a new one using automatic functionality in Word.
We are starting a project to advocate for and develop accessible thesis templates which we can roll out from September 2022.
Our ambition is that all theses are published to the institutional repository as fully accessible documents and that the students are supported to do this beyond the current training offer. This is being project managed, due to the interdisciplinary collaboration required between Information Services, Graduate and Researcher College, Student Support and Wellbeing, the Divisions and the Schools within these, Executive Group and the students themselves.
Keep an open mind when testing
We have found that testing the accessibility of digital works requested from our repository does work. However, we must make our own judgements otherwise we will fail to provide a fully accessible and meaningful format.
Learn and improve
There have been unanticipated benefits to taking a reactive approach to responding to accessibility requests. The limitations of testing are a good example of this and means that we can constantly improve our understanding and the service we offer.
We are lucky to have ‘accessibility-expert’ support from Ben and Nik (Student Support and Wellbeing). This network is vital to keep ahead of changes to the law, to develop our skills, and find solutions to new challenges. We’re also happy to answer any questions you may have about our process and have already worked with several institutions to do this.
If you would like to talk to us about this, or any of the other topics covered in our Accessibility and Inclusion News Blog Series, please contact the Research and Scholarly Communication Support Team.