Highlight text to annotate itX
The EPUB 3 specification is young. It was less than 2 months ago now, that the IDPF
membership voted unanimously to adopt EPUB 3 as their e-book format moving forward. And
this youth of the format has, with it, a lot of work that lies ahead of us. A specification
is merely a blueprint, and what remains now is to build the building base of that. In
this presentation I will talk about some of the steps that lie ahead of us, in particular,
pertaining to publishing and approaches to publishing in EPUB 3. I will also give you
an overview of the accessibility features of EPUB 3. So there will be a little bit of
technical details, but not too much.
It's important to know the potential of the format. If we look at the pure format from
a technical level, EPUB 3 is more or less outstanding in terms of the number of features
that it includes that pertain to accessibility. If we compare to EPUB 2, the predecessor,
there are magnitudes of difference in terms of accessibility features. In fact, if we
compare to the DAISY format, EPUB 3 even excels there in terms of accessibility features.
So, when we're talking about the DAISY Consortium merging with the IDPF on the format level
- this specification is the first step in that merger process - there has really been
no compromise from DAISY's perspective. It's rather been an enhancement and an improvement
to the DAISY standards that we are seeing coming into fruition here. And of course,
the merge has as its primary property the fact that EPUB 3 obviously is not a specialized
format but mainstream. So the excitement and happiness among the DAISY members around the
world is, I lack English words to describe the amount of excitement. These are very exciting
times and we have a lot of work ahead of us to make sure that EPUB 3 becomes a success
I am showing on the screen the cover of an O'Reilly book called "What is EPUB 3?" It's
written by a guy named Matt Garrish, who used to work for the Canadian National Institute
for the Blind, and also was the chief editor of the EPUB spec. This is available for free
on the O'Reilly website in EPUB and, *intentional cough*, PDF format. [Laughter] I went to the
site a week ago to check out the user reviews - as you may know, O'Reilly books all have
user reviews at the end of each page for the book - and this book had gotten top scores,
5 stars all the way through, at least a week ago. One of the readers said "I found this
book so good, so the first thing I did was to search the O'Reilly catalog for more titles
by Matt Garrish" - the bad news being that this is the only one, the good news being
that the next one that we will publish as a collaboration between IDPF and O'Reilly,
the next EPUB 3 related book published in Q1 of 2012 will be "EPUB 3 and Accessibility."
So as a token of IDPF's focus on accessibility, expect that book to be out in Q1 2012 and
I have a lot more detail that I will have time to go into here.
If that isn't enough good news, Matt Garrish is actually here with us today as well, any
questions or suggestions you may have, direct them to him and he will be able to help you
A quick look at the scope first. Print disabilities - what are they really? We are at the NFB
today, but the scope and the reach of accessibility technologies goes way beyond what many of
us think about when we think about accessible publications. We have individuals with cognitive
and information processing issues, this covers a multitude of conditions, the most common
being dyslexia. It is commonly said that approximately 1 out of 10 people have some form of dyslexia,
ranging from mild to severe.
Physical and neurological issues - dexterity, and so on, are also part of print disabilities.
And people with sight loss, from zero sight to a reduction of sight.
Also, let's talk about aging. There has been search in the European Union that 21% of people
over the age of 50 have some form of vision, hearing, or dexterity problems - that means
they need accessibility technologies in order to to consume information. 21% of everybody
over the age of 50. What does that mean? Well, let's look at the numbers. In 2010, the global
population over 50 and 21% of that means that today 300 million people are in need of some
kind of accessibility solutions for consuming information. And demographically we're an
aging population, right? The numbers are growing - in 2025, there will be 440 million, and
in 2050, 660 million. So if you meet anyone, a publisher or whoever, that tells you that
accessibility technologies and accessible publishing are a niche phenomenon for a small
group, I suggest you try to find a diplomatic way of telling them that they are gravely
So of course EPUB e-books, since their inception, come with a lot of promise for people with
print disabilities, the print book, the print paradigm being natively inaccessible for the
print disabilities that we are talking about. Over the years, there has been an evolution
and, as has been said before, things are getting better constantly. The problems of e-books
for print disabled users can be discussed in terms of the term "usability" which is
a general term defined in a multitude of ways. On Wikipedia, if you go there, you'll see
it defined as the ease of use and learnability of a human-made object. There's also actually
and ISO standard that defines usability as a measurable concept. I think this is an interesting
approach. So they're defining it as the extent to which a product can be used by a specified
user to achieve specified goals with effectiveness, efficiency, and satisfaction. That is something
that you can measure. Typically when you talk about usability you are thinking about tools
– software or hardware. And you can measure the usability, if you have two tools that
do the same thing, you can measure their usability objectively to tell which one is the most
useful in terms of effectiveness, efficacy, and satisfaction. So we're thinking about
tools, of course, reading devices for e-books are the heart of this discussion. But one
of the new things that we need to take into account as we move now into EPUB 3 and the
next new generation of e-books with many new features is that usability as a principle
also applies to e-book design. The previous generation, if you want to call them that,
of e-books, were basically static content that in one way or another tried to mimic
the print paradigm. But now as publishers are no longer looking at e-books as kind of
a second class citizen, but actually high-powered, very valuable products with a lot of new features
in them, this new generation of books will require us to look at the e-book design process
as something that must be approached very carefully and with usability concepts in mind.
Looking at e-book accessibility specifically – the ISO standard said, the extent to which
a product can be used by specified users to achieve specified goals. Of course in this
case specified users are different categories of people with print disabilities. But it's
important to note, also, and it's almost an insult perhaps, historically, to talk about
satisfaction as the consequences of inaccessible content for print disabled users. It's not
a reduction of satisfaction. It's the complete, sometimes, inability to view the content at
all. So it's a total catastrophe in terms of information access. But I think, as we
move forward, evaluating production tools and reading systems and so on, doing that
with usability concepts and principles in mind will be a good approach for us.
Another thing that needs to be repeated – the entire supply chain needs to be accessible.
We are beginning with the provisioning into phases – retail stores, online libraries,
and so on. There, the phases need to be accessible. The e-book itself, of course. The reading
system needs to be designed, have features that support this kind of usage. And finally,
also the assistive technology needs to be properly designed. That's something we take
for granted, that everything works as intended, but that's not always the case, right? We
have situations where the provisioning interface where the e-book and reading systems are great,
are perfect. But there's a bug or lack of updates and you still have inaccessible content.
Approaching the future, to get EPUB 3 to work, we need to keep this in mind. There are multiple
levels where things can go wrong.
In terms of accessibility guidelines, there's a huge amount of activity right now, more
entities producing e-book accessibility guidelines that I can count on my hands, I think. One
of the things that IDPF and DAISY will be doing is trying to start and coordinate this
so that there's no duplication of effort and we have some kind of centralized information
available to all stakeholders regarding accessible publishing. That said, I wanted to bring you
a point or two to one of these efforts by Ohio State University who published this recently
within the last month. A document called Accessibility Issues in e-books and E-Book Readers where
they go into detail in terms of functional criteria for e-book accessibility divided
by functional limitations, so all different categories of print disabilities they outline
both regarding e-book limitations and readers, the functional criteria. So that's a really
good overview and introduction in terms of understanding the variety in the print disabled
Let's start talking a little bit about the format and about a state of mind or an approach
that we can think about as we attack this problem. I'll assert that e-book accessibility,
and I'm talking again only about e-books, of the four steps in the chain, I'm only talking
about one of them. I'll assert that e-book accessibility is composed of two concepts:
the first is general sound offering practices. And with sound here I'm not talking about
audio, but healthiness. And the second one are what we can call superstructures that
are added explicitly for accessibility. And I'll show you what is what next. The good
news and the main point here is that general sound authoring practices will take you a
long way to where you need to be to do accessible content.
What are they? This was mentioned several times in the previous presentation as well.
The first and foremost sound authoring practice is about structure and semantics. As we know,
the publishing business comes from a paradigm where typography and visual layout are key.
It's kind of a transition for this business to understand the visual typographical layer
has a layer beneath it which is the actual code and the actual structure and semantics.
It is extremely important for accessibility and for the general usability for all users
to make sure that authoring tools can generate this kind of structure and semantics and also
of course that reading systems can read it. So comparing with DAISY and EPUB 2, EPUB 3
adds new elements and structures that didn't exist before, based on HTML5. There are new
things that we can do, better structure, better semantics that we can express, which of course
is very exciting to see that happening in mainstream markup language. For example, MathML
expressing mathematics is one of these new things we can use. So here's an example of
code, and I won't go into too much detail, but those of you who have dealt with markup
and documents will know what this is about. One of the things that EPUB 3 adds is what
DAISY had for many years, which is section to add proper top-level structure to the document
so you can wrap different chapters and parts in a project and markup using sections. There
is a number of other new developments as well for basic markup, such as the figure element
for figures with descriptions and so on. Another thing that EPUB 3 adds in terms of structure
and semantics is a way to do what we call semantic inflection. So, taking books as the
example, although EPUB 3 is not limited to books per se, there are book constructs that
people know and love that aren't really available in the basic HTML language. HTML is not defined
to represent books per se. So we add what we call domain-specific semantic inflection,
that's the complex term but it's actually not as complicated in reality. Take a note
apparatus as the example. So you have this book, and you're in a university class, and
there's this book written by some fanatic professor who spends 50% of the content on
notes. Your teacher tells you that you don't need to read the notes, they won't be on the
test. So what we can do with EPUB is to make sure the note apparatus, the link to the notes
and the notes themselves, are properly unambiguously marked up as such. This has many consequences.
If you are a sighted user and you are using a device with a limited screen, what you can
do here now is to simply ask the reading system to not show the notes and thus using the screen
real estate much better.
If I'm not a sighted user and I'm using a screen reader or synthesizer to render the
text in audio, this is also fundamentally important. It's something we've done in DAISY
for a decade, but for publishers in the mainstream it may be news, so it's important to mention.
So what I can do as a screen reader user now is simply to ask it to skip the notes, not
announce any links, not read the notes. And in terms of equality and equal opportunity
for everyone to get through university, having 50% of noise in this case removed from a reading
experience can be the deciding point for being able to take the exam.
Structure and semantics are very important and one of the main things that we need to
make sure are introduced into authoring tools in the mainstream. Without this, we don't
have the foundation to build the rest of the stuff on. Second in authoring practice, rich
navigation. EPUB 3 reshapes the navigation features that we had in EPUB 2 which was landed
from DAISY. The beloved NCX. What we've done in EPUB 3 is to upgrade that to an added number
of new features that will make it more appealing to publishers to use productively and also
included in ways that avoids duplication of data. I won't go into much detail there, but
of course what this navigation feature in EPUB lets you do is classic TOC navigation,
the ability to go to page boundaries in a source print book being represented digitally,
very important in a classroom situation of course. And also something that we call Landmarks,
which is a list of navigation points that let you go to specific points of interest
in the book – such as a glossary, or index, or the start of the text because everybody
wants to skip the front matter.
Rich navigation is authoring practice number 2, and to some extent it depends on the previous
authoring practice. If you have good structure in your document to begin with, you get this
more-or-less for free, it can be auto-generated. If you don't have bookmarks to begin with,
this will be a very poor navigation feature, or it will have to be hand-tweaked to work
by the publisher.
Those are the two basic fundaments, and this is really nothing about accessibility, this
is general usability, there is no consumer of e-books that will not benefit from this
being done correctly. Everyone benefits.
Superstructures – on top of this fundament, we add new things explicitly for accessibility
purposes. Worth mentioning and keeping in mind is that EPUB is based on web standards.
It's not a proprietary specialized format, it's using general and widely adopted web
standards. This means that there's a body of knowledge, lots of documentation and guidelines
on how to do the markup correctly already available, we get all of this for free. W3C
Web Accessibility Initiative, for example, are publishing the Web Content Accessibility
Guidelines, WCAG. What is stated in there applies to EPUB e-books directly. From a publisher
perspective, this is also good news of course that you don't need to look far to get the
expertise and the help you need in order to get this done. There is a huge amount of professionals
out there that mastered this and can help out to do it.
Examples of what the WCAG guidelines include regarding accessibility – for example, to
use a track element when you include videos, to have captions in audio descriptions – again,
not specifically accessibility related, everybody wants captions and descriptions are sometimes
good. If you have images in your content, of course providing alternate text is basic
stuff today. Another thing in terms of images is the DIAGRAM project funded by the Department
of Education here in the U.S. to provide an ecosystem for access to descriptions of graphical
content. This might be mentioned again later today, I believe, but one of the problems
we're having currently in HTML5 is that long externalized descriptions of complicated graphs
are not supported at this point. The longdesc attribute that we had in previous versions
of HTML has been removed. So DIAGRAM Center is approaching this problem and trying to
build a new ecosystem where publishers can use online repositories to some extent even
crowdsourced in terms of writing the descriptions and being able to share and include such in
One of perhaps the most controversial features of EPUB 3 is scripting. When we're talking
about the next generation of e-books, interactivity is perhaps one of the most anticipated features.
E-books are no longer static, passive pieces of content but things that you can interact
with in one way or another, spanning from doing tests in the classroom to more perhaps
casual features such as images made just to look cute. Here's one of the second big challenges
– to make sure that we get a good publishing practice and a good publishing culture moving
forward to add the scripting in a way that doesn't have negative effects on print-disabled
users. One of the technologies to use there is ARIA from W3C. This is one of the basic
building blocks to make scripted content accessible. Next feature in EPUB 3 for accessibility is
Media Overlays as we call them. This is the re-enactment of the classic DAISY text and
audio synchronization feature. So you can now have the full text of the book and pre-recorded
audio hopefully by a human, coexisting in the book and synchronized so that you can
read them together. What we've seen is that there's been a huge interest in this and again
not particularly for accessibility purposes but as a way of making enhanced e-books. For
kids learning to read or for children's books in general, having this feature has proven
to be really cool and Apple iBooks announced support for Media Overlays back in spring,
even before EPUB 3, so as many of you probably already know, you can already enjoy some of
these books from the iBooks store.
Next, Text-to-Speech. What to say about this? In the professional usage areas of e-books,
it's perhaps one of the most important features. We already saw text-to-speech demos in the
previous speech here. One of the problems we're having with TTS is that while it may
sound good for your bread and butter texts, as you get into advanced content such as university
literature or you have many specialized terms and so on, even the best and most expensive
synthesizers begin to fail increasingly as the content becomes more advances. So it's
ironic that one of the primary use cases for TTS as opposed to pre-recorded audio tends
to be among professionals and students, because you want to have the ability to control speed
and also control, as we saw, walking among lines and repeating words and all that. So
it's an extremely important tool. Where it would be used the most, it's also where TTS
begins to fail. So what EPUB has added is the ability for the publications to include
instructions that help TTS engines to pronounce difficult stuff better. So there's pronunciation
lexicons and other kinds of instructions that you can include in the book to make sure that
the difficult passages are handled okay. This is advanced stuff and it's not something that
is expected for general publishers to master manually, you need computer linguistics skills
that exceed most mortals. What we are expecting to see here is basically specialized tools
– imagine that you are able to submit your EPUB onto some service and then get it back
with this stuff added.
Another point is reusability. Pronunciation lexicons, for example, can exist as global
masters in a publishing house and then as you produce an EPUB you pick out the lexicons
that are relevant for the publication. So lexicons are re-used to create new ones for
every book you do.
Metadata – this is pretty new, and also extremely important moving forward to get
this working right. This is another place where we need to work with publishers and
retailers and libraries. The most recent piece of good news is ONIX Codelist 196. I'm not
sponsored by ONIX, so in terms of neutrality let me say that there are other popular metadata
standards out there. ONIX seems to be first in adding specific metadata codes for accessibility
features. Codelist 196 was ratified in October this year by the ONIX steering committee.
So what can you do with this? ONIX metadata is typically not embedded in a book. It's
metadata that sits in the retail chain and it is also dynamically updated. So it's different
from metadata inside a book which is the static stuff like title and author.
I'm showing a slide here with some of the features that you can use. You can basically
specify which accessibility features are available in a book: Can you navigate via TOC? Is there
an index you can navigate by? Is the reading order chronological? Is there alt text for
the images? Is there a full alternative descriptive text for graphical content? Is the math accessible?
Is the chemistry accessible? Is there page numbering that is equivalent to the print
source? Does it have Media Overlays? Basically, what this enables is for consumers to be able
to match their preference profile, somehow submit that into the library or retail interface
and make sure that the books offered to you meet your expectations in terms of accessibility
features. This is an extremely vital part, this way our users will be able to know what
they are getting beforehand.
So ONIX Codelist 196 is not a part of EPUB 3, but a separate effort by an organization
that owns ONIX.
One final slide that I added jus this morning – fixed layout is one other, talking about
things we need to focus on moving forward. Typically, e-books are what we call dynamically
reflowed. So depending on your device the text basically reflows on screen to utilize
the screen real estate. Fixed layout publications are pre-paginated with static content. And
for reasons that we're still trying to understand, there's a current trend, at least in the U.S.,
to use EPUB fixed layout publications, lots of it. And there are problems in this for
publishers right now because EPUB still needs to standardize metadata around fixed layout.
There's also a potential accessibility trap, because fixed layout content tends to be graphically
oriented. I'm showing here an example of a double-page spread from a magazine where there's
a picture of some green stuff in a glass, supposed to look like a tasty drink, but I'm
not sure if it does so. The picture takes up the entire screen, and then there's a recipe
superimposed on top of the picture. This could potentially be, and depending on how this
EPUB was made, this could potentially be completely inaccessible, meaning that also the text here
is actually not properly marked up text but just part of the image. It could also be 100%
accessible, if the publisher knew what they were doing when they were doing this, in other
words using other constructs than mere images to provide this kind of info. So that's one
of the focus areas of IDPF and also in terms of our guidelines for accessible publishing
I think we need to really be focused on this area – fixed layout books.
We've been talking about a number of terms. In terms of accessibility, we've said that
there's a basic fundament which benefits all users which is the need for good structure
and semantics in the markup. We've said that EPUB 3 adds many new elements and ways to
do this that excel over both DAISY and previous versions of EPUB. We've talked about rich
navigation as a fundamental feature of any e-book, as a necessity and requirement for
all users. That's the fundament and getting those two done will take you almost all of
the way to where you need to be to have a book that is accessible to all and usable
by all. There's a number of additional technologies in EPUB 3 that you can use. We have the web
accessibility techniques that everybody should use without even thinking about them, we have
Media Overlays, ARIA for scripted content, Text-to-Speech enhancements, and the ONIX
So, to begin to summarize, what we are trying to tell publishers, what we will enable publishers
to do is design books with usability in mind, target all users, also then sound authoring
practices. The basic general good behavior in terms of document production will often
be all you need to get where you want to be. Scripted content needs extra attention. This
is one of the areas where we need to focus our efforts in terms of guidelines to publishers.
Media overlays sees the bridging of the audio business and the e-book business, suddenly
it is possible for publishers to make the audio book and the e-book one and the same
and let the user decide whether to read it as audio or as text or as both, synchronized.
What we are saying to publishers is, add media overlays because everybody would probably
at least segments of users are interested in this. And by doing that, you get accessibility
benefits for free. Regarding TTS, again advanced technology but extremely crucial in professional
and academic environments so our recommendation to publishers is that for text books in particular,
include the new EPUB 3 TTS features which in this case in the coming years will consist
of looking for a service that allows you to do this for you. And also then, to publishers,
in the case that you're based on ONIX in retail, make use of the new ONIX code list to declare
the nature of your book in terms of accessibility features.
In terms of publishers again and the recommendations on a general conceptual principle level, here's
a few, shall we call them, mantras or recommendations again: First of all, and we talked about the
typographical worldview of traditional publishers – there's nothing wrong with that, but as
you move into the new generation of e-books enhanced with rich new features, you need
to realize that e-books are not typographical only. There are two levels to a book – the
visual layer and the underlying layer, and the publisher who wants to be forward looking
and progressive and produce really good content needs to take control over that code. There
are problems in publishing tools with regard to this, and that's one of the main areas
that we need to focus on fixing.
Second, as I said before, web accessibility and design principles are another basic fundament.
All the knowledge, all the professionals needed are already out there, so this is not a difficult
part to attain. Also, in terms of the next generation e-books, a recommendation to publishers
is to try to interact with users early on in the production chain. We're going to see
over the coming years fantastically interactive e-books with new features that we haven't
managed to dream of. The usability of these features, again, the effectiveness, efficacy,
and satisfaction that result from it is going to need to take some time to mature. We're
going to see good examples and I'm sure we'll see some bad examples again. Evaluate content
early on in production chains. Set up a focus group. Let that focus group include people
with print disabilities. Have them pre-evaluate your content and recommend how to deal with
it. That's going to improve the quality of your product for everyone for sure.
Also, in terms of tooling, try to avoid single vendor lock-in. Keep an eye open, continuously
evaluate new tools that may help your publishing flow get better both in terms of economy but
also in terms of quality of the product.
And finally, validation. Somebody who pays my salary asked me to put this in here, so
I need to mention this in every presentation we do, apparently. And I'm also the guy who
writes the code for this tool, so I should've thought of this myself perhaps. Validating
content is extremely important. There is a tool for EPUB called EPUBCheck, which already
supports EPUB 3 enough that you can use it and trust yourself to get good results. Validate
your content – never submit content that doesn't validate. Make sure it's valid. What
EPUBCheck does not do is specific accessibility checks. So you can have an EPUB that is 100%
valid according to this tool but still may lack some fundamental accessibility features
such as good descriptions for images. A software tool cannot tell you whether a piece of text
is bogus or actually correct. So there is also in terms of validity and quality control,
of course there are interactive processes as well that need to be used. Machines can't
do everything for us, unfortunately.
Okay, so final slide and summary. Accessible e-books are not for a small niche. We talked
about how 1 in 10 have dyslexia, and we saw the numbers of the aging population in the
world. It's very important to remember, this is not a niche at all. This is something that
you should consider a normal part of your production flow. EPUB 3, as I said as well,
includes on the format level enough features to make it a state-of-the-art bleeding edge
foundation for good accessible publishing.
And finally, in terms of production tools, I think one of the most critical things that
needs to happen over the coming year is that we see an improvement in existing publishing
tools to include the features that we've talked about. In fact, there is really nothing to
prevent the big authoring tools we have out there for publishers such as InDesign and
Quark, there is nothing to prevent these tools from including all this stuff for publishers
without the publisher really having to worry or think about it. The high-end publishing
tools can do this for us, but at this time they do not. So one of the recommendations
and perhaps one of the things RNIB can help out with as well is work with publishers to
make this happen, to integrate these features under the hood in the existing visual tools
out there. There's also this notion of accessibility checkers, for those of you who use Microsoft
Word for example know that the accessibility checker in there is now a default part of
your distribution. So it's a little add-on tool that you can invoke and go through the
document and tell you if there are risks for inaccessible areas. And having such an add-on
integrated in the big publisher tools would of course be fantastic as well.