EPUB and Accessibility

This is an exciting period for content authoring and publishing in higher education. A paradigm is emerging of tools, standards, practices, and processes that promise to accelerate and enrich the impactful learning experiences designed and delivered by our institutions. Any paradigm shift raises both opportunities and challenges, and I want to address one particular issue in this article related to the EPUB content format standard.

Now in its third generation, EPUB is a semantic document format developed by the International Digital Publishing Forum (IDPF). It combines all of the main tools of the modern web – HTML5, CSS3 for styling, Javascript for interaction – and extends them with an expressive authoring vocabulary for digital publishing. The tools needed to produce an EPUB are more affordable and easier to use than ever. Many desktop applications export to EPUB directly. EPUB reading systems, for their part, are maturing to leverage all of the features of EPUB. Publishers, academic institutions, faculty and the open educational resource community are embracing the standard. With EPUB, you can design content to be as media-centric and interactive as any contemporary website.

The aspiration to produce media-rich, interactive educational content for higher education is perhaps nowhere better expressed than in “EPUB for Education.” Formally called “EDUPUB”, “EPUB for Education” is a content framework that combines IDPF’s EPUB standard with IMS Global’s LTI, QTI, and Caliper interoperability standards. These standards – used for tools integration, question-test interoperability, and learning analytics, respectively – equip the EPUB author with all of the basic tools to produce complex, media-rich educational applications and courseware (not just a traditional book) fully integrated into tool ecosystem. Indeed, publishers (whether in the academy or otherwise) can produce totally self-contained learning experiences, assessments, grading rubrics and real-time data.

As EPUB represents the future of digital publishing, so “EPUB for Education” marks the future of digital publishing in higher education. But in the promises of the EPUB/EDUPUB content format lie a number of tensions with the mission of our institutions. One of them, for example, has to do with the mission of our Universities to provide accessible learning to every student regardless of their sensory or cognitive abilities or disabilities.

Suppose you are student with a print disability – such as low-vision, blindness, dyslexia, etc. – in an introductory biology class. The textbook required for the course is choc full of figures of biological processes, data tables, and complex graphics. The prose may be structured with exercises and side-matter not part of the main text. How do you read those figures, tables, and graphics? How can the information be organized into a format that you can use in combination with technology so that you can read it? How, in short, can the information be made accessible to you? Well, blind students will rely on assistive technologies (ATs) to provide accessible (e.g., audible) interfaces to computers and files. But not all content formats are equally accessible to ATs. The solution is that University staff will need to produce an alternate version of the content in a format that is accessible to users through their ATs.

Producing accessible content from digital sources is a challenge for universities. Some content formats are easier to start from than others. Adobe’s Portable Document Format is the world’s most popular document format. While “born-digital,” the content format is portable but infrequently is the content accessible to most AT users. In fact, born-digital content in any format still usually means born-inaccessible.

Producing alternative, accessible versions of content is laborious. Some of it can be automated, but the majority of the work requires the careful inspection by trained staff. The manual labor is part science, part art. How, for example, do you describe a photograph to a blind student? The problem can be tricky.

What can be done to make it easier and more efficient for our institutions to accommodate the content needs of their students?

One part of the solution is collaborating with publishers. Unizin is partnering with publishers to make it easier for our institutions to quickly respond to student needs by creating accessible version of work in advance of formal publisher approval. We are also inspired by models like BookShare, and believe that institutions should collaborate in producing accessible versions of content rather than, as is the case today, duplicating efforts.

Another part of the solution will lie in the content format standards themselves. It is here that the promise and challenges of EPUB and EDUPUB come into focus.

In theory, the “EPUB for Education” standard enables the production of born-accessible content. But the bar to clear to achieving born-accessible content is high. Authoring tools, practices, workflows, and production processes, as well as reading and content delivery systems, all need to mature to respond to the diverse needs of students. Conformance processes for “baseline accessibility,” such as the one presently proposed by IMS Global and IDPF, are an excellent start. But there will likely always be the need for our institutions to produce derivative versions of content optimized for accessibility.

One key question is whether the EDUPUB format will, in practice, make it easier or harder for institutions and disability student services offices to create derivative, accessible content. There is reason to stay vigilant on this topic.

As I noted above, EDUPUB is a collection of four standards – EPUB (content format), QTI (question/assessments), LTI (tool interoperability) and Caliper (learning analytics). The EDUPUB format encourages the creation of rich, complex, self-contained learning experiences with their own assessments, tools integrations, and learning data independent of any particular reading system. The richer the objects, the more complex it is to untangle content and unbundle its different components so that an accessibility expert can make them accessible to a student. For example, will content authors bundle their EPUBs in ways that make it easy to produce accessible versions of rich media objects and assessments? What about interactions with web- applications embedded in the EDUPUBs? Unbundling an ED fact will likely pose enormous challenges to disability student services offices who must produce accessible versions of increasingly intertwined, self-contained learning experiences.

We believe that a simple set of conceptual distinctions would facilitate the important work of making content accessible. The concepts refer to a different aspect of educational content. We believe it might be useful to apply them in the way EDUPUB content is authored, marked-up, packaged, and distributed:

1. Media assets – the set of text, animations, videos, photos, images, figures, audio, and other media assets that, collectively, make up a single EDUPUB.

2. Instructional design – the design structure that purposefully integrates media assets into an intended learning experience.

3. Reading system – the application that consumes EDUPUB content and, like a web browser, delivers the learning experience to students.

When accessibility offices at our institutions produce an accessible version of content, they seek to preserve the instructional design aspect of content and create alternative formats of the media assets so that they work with the preferred reading system of the students they serve. Being able to unbundle the media assets components from the instructional design structure of an EPUB is key to enabling assistive technologies (i.e., reading systems) to deliver an accessible learning experience to the student.

Authoring tools for EPUB/EDUPUB content will need to make it easy for authors to encode their media assets separately from the descriptions of instructional design. The concepts and challenges above are not exclusive to accessibility. The re-use and portability of content assets from one instructional design context to another – a core goal of the OER community – requires the same kind of “debundling” described above. So does the continuous improvement of instructionally designed experiences delivered and measured in our reading systems.

It is good design practice to distinguish design intent from the media assets that facilitate it, and to distinguish both from the reading systems that deliver and measure a learning experience.


Etienne Pelaprat, Unizin Director of Product Management