DCL Learning Series

Marianne Calilhanna

Hello, and welcome to the DCL Learning Series. Today's webinar is titled “The European Accessibility Act: Navigating the Nuances Before Time Runs Out.” My name is Marianne Calilhanna and I'm Vice President of Marketing here at Data Conversion Laboratory. Before we begin, I do want to let everyone know that this is being recorded and it will be available in the on-demand section of our website at dataconversionlaboratory.com. We'll absolutely save time at the end of our program to answer any questions. And do feel free to submit any questions or comments as they come to mind, and you can do that via the question dialogue box. So, quickly before we begin, I do want to take a moment to introduce my company, Data Conversion Laboratory, or DCL, as we are known. We are the industry-leading XML conversion provider. DCL's core mission is transforming content and data into structured formats. We believe that well-structured content is fundamental to fostering innovation. And the services presented on this slide speak to that mission with DCL's solutions like semantic enrichment, entity extraction, migrations to content management systems and tools for content analysis. And the bottom line: if you have complex content and data challenges, we can help.

But of course, the reason we are all gathered together today is accessibility. And in the content world accessibility starts with structure. I am so delighted to introduce today's speakers. Mark Gross and Bill Kasdorf have spent their careers in the content software and technology industries. Mark Gross is president and founder of DCL. He established Data Conversion Laboratory 43 years ago and is an industry-leading leader in structuring content and digitization best practices. Bill Kasdorf is principal of Kasdorf & Associates and a founding partner with Publishing Technology Partners. He's an expert on accessibility, XML, HTML, EPUB modeling and specification, standards alignment, and publishing workflows. Thank you both so much for sharing your knowledge and expertise with us today.

Before we begin and before I turn it over to your conversation, I would like to just quickly make sure we're all on the same page with a broad definition of the EAA. So, the EAA is a law that aims to make products and services more accessible to people with disabilities across the European Union. It sets accessibility requirements for a range of goods and services, including computers, smartphones, banking services, eBooks, and online shopping. Ensuring they can be used by everyone, regardless of their abilities. The act also promotes consistency and accessibility standards across EU member states. And I urge everyone to take the time to read through the EAA. It comprises 11 chapters and six annexes. And my colleague is going to push the URL to the EAA out to everyone right now.

4:03

These are key deadlines and dates that I pulled out of the EAA. And the 28th of June 2025 is the primary deadline for businesses and organizations to ensure that their products and services comply with the EAA's accessibility requirements. The details around these dates and associated compliance is what Bill and Mark are going to discuss. So with that, Mark, Bill, let's get into some of the complexities and nuances of the EAA. Over to you.

Mark Gross

Well, certainly there's a large document involved in it, the EAA document we're talking about over here. And this is a section that Marianne's put out that talks, just gives you a small synopsis, a small snippet of it. And the document really talks about a lot of other things than publishing and books and things that talks about architecture and signs and all kinds, but really focused on really a small piece of it. And I think, I guess where I would start, and I'd ask Bill to comment on it is just, when does this take effect? When do people have to start worrying about the EAA regulations here? And also talk about what is it? Is it a regulation? Is it a guideline? What is it that we're looking at right now?

Bill Kasdorf

Well, that's certainly fundamental, Mark. Thank you for that first question. And to answer one of your three or four questions that are involved in that question, is that if you sell products or services in the EU, you should have been worrying about this before this webinar. You're getting a late start compared to most publishers who have been preoccupied with this almost since it was announced in the first place. But as Marianne said, it takes effect June 9th next year, so that's less than a year. I'll get back to that in a second. A fundamental thing that I always say when I start talking about the EAA, it's a common disclaimer, which is I am not a lawyer. And the reason that that's so important is that one of – I do a lot of work in this area, whoops, I'm getting some feedback from someplace. I do a lot of work in this area. I'd say probably at least the majority of my consulting, all my consulting in some dimension addresses accessibility. Because I do a lot of content modeling and workflow work, et cetera.

But I'd say probably at least 50% of my work is just analyzing publishers' books and journals for accessibility and for whether they're going to be compliant or not. So, it's something that lots of people are paying a lot of attention to. But the first thing I say to them is “I can give you some guidance, but you need to consult your legal counsel for the absolute concrete advice from a legal point of view.” Because the document is a very complex legal document and you almost have to be a lawyer to read it, to really understand it well.

8:03

I've read it in depth, and when you first encounter the directive, most people are like “OMG, this is so confusing. It's talking about disproportionate burden. What do they mean by that? It's talking about micro enterprises. What's a micro enterprise?, et cetera. Why didn't they say what it is?” Well, you need to read the whole darn thing, because there's pages and pages at the very end of definitions, and they clearly define these things. So in that making you wonder, micro enterprises are the simplest criterion is fewer than ten people, so most people attending this webinar do not work for a micro enterprise. Micro enterprises are exempt. Then it refers to SMEs throughout. Well, in my world that's subject matter experts, right? Well, SMEs are small and medium enterprises in the EAA. And now we're up to 250 employees or members and two million euros of turnover.

Anyway, the point is, what probably initially will mystify you is actually spelled out, but it may be spelled out in a whole different part of the document from where you encountered it. So that's one reason why people get confused by this. But lawyers are used to sorting those kinds of things out. But there really are some ambiguities, which I'm going to get to in a minute. But to get finally back to Mark's question in terms of the "What is this thing?" One of the things I also always stress with people is, and that people often don't clearly distinguish between our regulations, guidelines, and specifications. Those are three different things. EAA is a regulation. So it says you're required to do these things by this date. It doesn't actually tell you how to do them, it just says you have to do them, right? It's not a specification, it's a requirement.

Now, one thing that it does specify is that you have to align with WCAG AA, which is there are three levels to WCAG. Level A is the bare minimum and nobody in the accessibility world would accept that as being sufficient. AA is almost universally understood to be that's what you're really expected to do to be considered accessible. AAA as well. If you really want to go all out, you should do these things too. Those would be great, but they're kind of the icing on the cake. And most requirements don't expect AAA conformance, but they do expect AA conformance. But WCAG, the G in WCAG is guidelines, Web Content Accessibility Guidelines.

And it actually is not a specification either. It doesn't tell you how to code your files, and we're talking about websites or books or journals. It provides what are called success criteria. And the example I usually give to people on that take a website example, is most websites will have a form field to get filled out. So, a user of assistive technology needs to be able to A, find that form field when they can't see it on the screen like most of us can do. Once they found it, they have to know what it is, what are they supposed to fill in there? So that's another thing that's called the success criteria. I can find this, I know what to do. The third thing that you have to be able to do is know that you've successfully done it.

12:01

So, there's a focus issue there too. I don't need to get into any more of those kind of technical details. But the point is that it doesn't say how to code that form field, that's specified by the HTML and an aspect in the HTML world called ARIA markup that is focused on enhancing HTML for accessibility. And so it's actually quite simple. You need, one thing you need is something like an ARIA label or title that just says “Here's what this thing is.”

So, in terms of a specification, particularly for books, really the one you need to focus on is EPUB Accessibility 1.1. It's a formal final recommendation by the W3C. EPUB has not had that kind of formal adoption and process until EPUB 3.3. Accessibility 1.1 is a component of the EPUB 3.3 recommendation, and that should be your guide, really. I can give you some guidance, I can give you some recommendations, et cetera, but really that's the thing to do. And I happen to have had a glance at the registration list, and I happen to know that I've got some good friends that are signed up to take this seminar, or this webinar, that are really experts in this, right? For example, one of the countries in the EU that's really been in the forefront of getting all this right and helping this stuff move forward is Italy, and there's an organization called AIE, which is part of the Italian Publishers Association that focuses on accessibility, and Cristina, for example, Cristina Mussinelli, has been involved with EPUB development as long as I have: over ten years, probably. And Elisa's on the call. I'm working with her in a project with DAISY to work on image description, guidance, et cetera.

So anyway, I encourage people on this webinar, if I'm saying something that you don't understand, and particularly for people like Elisa and Cristina, I don't know if Gregorio is on or not, but there are others too that I know are very expert in this. Put it in the chat because if I'm getting something wrong, you may know better than I do. You're European and you've developed this. And one reason I'm going off on what's, it may seem like a tangent, is that some of those folks, Gregorio specifically being one of them, as the EPUB Accessibility 1.1 Spec was being finalized, literally went through it with a fine-tooth comb and made sure that how did every aspect of that specification align with WCAG AA and therefore with EAA? And so that work was done. It's been documented. You can get it, it's a W3C document. There are a whole bunch of other very useful W3C documents, accessibility, metadata guidance, all that kind of stuff. But anyway, that's where you actually get the detail about here's exactly what to do. So, I probably rambled on too long. Mark, do you want me to get into that? When do you have to do this question, because that's going to be a bigger discussion.

Mark Gross

Well, I just wanted to go back up a little bit on what this applies to, because what I saw was that the EAA, the document refers to eBook, but it seems that eBook is really in a – first of all, what do we mean by eBook? Because I think it also talks about other e-documents rather than that.

16:00

And following questions going to be talked about the EPUB. But there's all these documents that are in JATS and in BITS and other formats that have been used for millions and millions of pages. How does that react to this? First of all, what are we talking about here that reacts to, and what about all those other documents that have been created over the years?

Bill Kasdorf

Okay, great. Those are excellent questions, Mark, and I'm going to take them in reverse order, because one of the things I always stress about JATS and BITS XML is that they are not rendering formats. In other words, they're not intended to get delivered to an end user. They structure and in minute detail describe the content so that it can be rendered in any number of different ways for an end user. But what are these things in your content? So, JATS and BITS are precursors, but there are ways to do your JATS and BITS to make it more accessible. I had the pleasure of being in what's called a subgroup of the JATS for our working group. It's a nice working group that develops specifications of how best to use – or recommendations for how best to use JATS for various aspects. Because JATS is giant, it's deliberately accommodating. It lets you do lots of different things in lots of different ways. Well, to accomplish this particular thing, JATS for ours says “Okay, here's our recommendation for how to use JATS to get this right.”

The one I participated in was the accessibility subgroup, we sent our recommendations in several months ago. And one thing that people need to know about that work that is we specifically did not recommend anything that required a change to the current JATS, which means that accessibility recommendation you can do right now with the way JATS is specified right now. Because it evolves over time. Mark knows that better than anybody because they're in the business of it staying up to date with JATS. And by the way, different hosts, for example, have their own specification for JATS, Silverchair JATS is not the same as Atypon JATS, right? So anyway, there's guidance there. But so getting back to the earlier part of your question, Mark, is that those questions, there's two there that are sources of what I consider great ambiguity. And one reason that I say it the way I'm saying it is that one thing that I think is dangerous is that I think there are a lot of people who are quite sure about things that I'm not sure of. I don't agree that such and such a thing is absolutely settled. And I'll say that most of the people that I look up to in the accessibility world tend to take that same position. George Kerscher being one of them, Richard Orme, head of, CEO of DAISY is another one.

An early promo for this event made the comment that the backlist has a five-year extension. He flipped out. He said “Why are they saying that?” Well, most people think that that's a settled issue. George doesn't think so, I don't think so, Richard doesn't think so. And the reason is that, well, there's a couple of reasons, but the bottom line is, we're talking about twenty-seven different countries,

20:00

and the EU has agreed to align with EAA. So even though it's not EU anymore, you need to conform to this for the UK as well. But each country still has to document how is it going to assess whether you're accessible or not? And only, well, he mentioned two, I think it was Italy and France have articulated that, and they've said that they're going to provide that extension for backlist. That leaves twenty-five others, twenty-six if you're counting the EU. Here's why I think that's not settled, not carved in stone yet, and why you need a lawyer. In a sense we need to wait until we hear from all the rest of the countries to make sure, because this is really complicated because of all the different countries that may not implement it exactly the same. And somebody may raise an issue to say “Wait a minute, I don't think this makes sense.”

That could upset the apple cart, and I don't want to throw a panic at you. I have a more consoling approach after I get finished with this comment that will let your blood pressure go back down a bit. So yes, there's the fundamental thing that everybody has been saying to justify why you have five years to make your backlist accessible is that the EAA considers EPUBs or considers eBooks, I'm sorry, it doesn't use the term EPUBs, it's eBooks, considers eBooks to be services, not products. Well, on the face of it, if you tell that to me just straight out like that, my reaction is “What planet are you on? eBooks are products. It's the delivery systems that are the services.” And they say “Oh, well, it's right there in the spec. It's right there in the regulation,” et cetera. Well, notice what DCL has on the screen right now. This is a really important page in the directive. The following products, blah, blah, blah, blah, blah, are affected. And then the following services, well, eReaders are a product, that means your Kindle, for example, is a product, so that's in the first category, but then it says eBooks and dedicated software.

It doesn't just say eBooks, it says eBooks and dedicated software. And if you read the regulation in more detail and more completely, it specifically uses the term eBooks, the – what exactly is the term? It basically says it's the process of delivering that eBook to a user, that's a service, eBooks and the whole delivery mechanism around it. Well, I think that opens the door for some lawyer someplace, particularly somebody that's a strict accessibility advocate to say, yeah, okay. And it makes intuitive sense that the aggregators and retailers of eBooks have to be able to stay in business for at least five years to allow this to happen. Why is that? First of all, they can't change the eBooks that they've got. The publishers have to do that. They have no ability to alter the files that they've got that they're distributing to make them accessible.

24:03

They literally can't do that. So it makes total sense to let them off the hook beyond next June, whether somebody's going to come along and say, or some one country is going to come along and say “No, we're not going to put Kobo out of business. We're not going to put Apple and Amazon out of business and make them shut down or make them deplete their inventory of 99% of the books they've got in there. But what we are going to say is that the publishers are supposed to be fixing those books, making those accessible.” So, like I said, I don't want to induce panic. I just want to make sure that I'm being cautious. And the reason is I'm a consultant. I'm responsible for advising my clients, and I'm very serious about not advising them on something that maybe is pretty true, I'm kind of pretty sure it's true. It's like that's not good enough for me. It either has to be true. I have to know it's true or I have to qualify it. Because one thing I cannot do as a consultant is advise somebody to do something that then gets them in trouble down the line. It's like “You've told us to do that, Bill, and now they're telling me that's not true.” So, I'm very cautious. I would say 99% of the people I work with are convinced that you have the five-year extension.

And by the way, there's a difference between an extension and an exemption. You're not exempt from making those books accessible. You're just given some more time to deal with all that backlist. Let's hope that that is what happens when this is all implemented and next June rolls around and nobody gets thrown in the hoosegow because their backlist isn't all accessible. For one reason is it's literally impossible for most publishers to do that. The thing that I think I stress more often than not in these conversations is I'm not going to tell you that I'm absolutely sure that you're going to get that five-year extension. You definitely have to do it for frontlist. We have to get them accessible. You may have an extension on the backlist, but I can tell you, now, this may be something of a self-selected sample, but every publisher I deal with, and I interact with lots of publishers on all these industry committees and things. So they're not clients of mine necessarily, but I know what they're doing. Every one of them is really working hard to address this.

And so what they're doing is they're diving into the backlist and doing what they can to get it accessible. Most of them are taking a strategy of looking at their, it's not the date of the book, it's the market for the book that is guiding them. They're finding which backlist books are most important to make sure are saleable, and they're getting those accessible first and then working their way back. And so, if they're doing everything they can, you really can't ask more than that. So calm down and just do what you can. Calm down, but just do what you can to get your backlist accessible. But in the meantime, make sure your frontlist gets accessible. And one other thing, and I'll let you talk again, Mark, sorry to be monopolizing the conversation.

Mark Gross

I can shut the microphone off. [Laughs]

Bill Kasdorf

Yeah, you can cut me off. Yeah, you can cut me off. So, most publishers that I work with, smaller university presses are the exception. But most publishers, other than the micro enterprises, maybe a small publisher that really is less, fewer than 10 people on their staff, they don't actually create the EPUBs. They have a vendor that does that work. They have a vendor that does the pre-press work,

28:00

and they probably send that, DCL creates the JATS, and then some types that are creates the PDF and the EPUBs ideally from that JATS, because now they've got good structure to work with that. That would be the ideal workflow.

Mark Gross

They go straight from that.

Bill Kasdorf

The vendors should know how to do this stuff. I'll have to say, I see a lot of crap EPUBs from vendors that I thought were pretty darn good vendors. And when I come to find out, you'll enjoy this little tidbit, Mark, and the audience too. I had one recently. It's very good publisher, a university press, one of the larger university presses. I know they work with good vendors. And one of their books that they sent me to analyze, an EPUB, had no semantic markup whatsoever. Everything was just <p>s and <div>s, no heading markup, no list markup, nothing. All that was just handled by class attributes, which is appearance related. That's not the semantics that assistive technology uses. Plus their metadata was garbage. What they were doing is putting in the same boilerplate metadata in every book. So one of the books they said had structural navigation, it had image descriptions and it had MathML. But if you look at the book, there are no headings in the book. There are no images in the book, and there's no math in the book. So that metadata is just plain wrong.

And that's actually deceptive to the people with disabilities, because they're looking, somebody looking for that book may say “Well, what am I doing wrong? I can't find the images, I can't find the math because it's supposed to be in here.” So, you just need to get that right. But in that case, really the vendor shouldn't be making that mistake. So I then, the publisher didn't identify the vendors of the EPUBs that they sent me. But after getting my report, they said “Okay, Bill, we want you to talk to these three vendors that are our three main ones.” And the one that did that book said “Oh, well, first of all, that's an EPUB2. And we did it about seven or eight years ago,” and secondly, well, I'll give you a different example from a different client also, as it happens, a big university press, they were using two vendors, both of whom I know really well, one of which is actually one of the first vendors that Benetech hired to test EPUBs for conformance to their global certified accessible program. So they absolutely know how to get it right, but they didn't, these EPUBs that they sent me were terrible.

And it's like, well, I know the guy that's the main architect behind all of that. I called him up, I said “What the heck's going on? You guys know how to do this?” And he said “Oh, well, we've been working with that client for fifteen years and we're using the spec that you wrote for them twelve years ago. They never told us to do anything different, so we just keep doing it to their spec.” One thing that publishers need to do is revisit your specs, for Pete's sake. Your vendor might be just cranking the books out like they have been. And if you haven't said “These aren't what we need anymore, we need you to do this,” their instinct and practices just keep doing what they're doing. And so anyway.

Mark Gross

What you're saying is, you need to get the order right. I mean, we want these books to be accessible so the vendor can do that. You were saying –

Bill Kasdorf

EPUB accessibility 1.1, just tell them that, they should know how to do that.

Mark Gross

Well, yes, but you have to say that. Nothing is implied, right?

32:02

Bill Kasdorf

Right.

Mark Gross

What you said, you can't say anything until you're sure is really an issue over here because this isn't really been out yet. This is the same issue as a webinar we did probably six or seven months ago about AI and copyright and –

Bill Kasdorf

Oh, I remember that webinar. Yeah, I attended that one.

Mark Gross

It's the same issue. The answer to almost everything was “This is what we think should be there, but we don't know because there is no case law yet.” And this also has been out there. There's no case law. No. So back to your earlier comment, I don't think anybody's going to the on go on this, but they probably, they may be told “You can't sell this until you make it accessible.” I mean, that may be the kind of thing that may be happening along the way.

Bill Kasdorf

But how are they going to judge what's accessible and what's not? They're not going to individually look at the metadata on every book.

Mark Gross

So let's talk about that.

Bill Kasdorf

Thousands of backlist books.

Mark Gross

What do we think – like you mentioned already that it doesn't have titles in the book. What are the other things that people need to worry about in their list or in their backlist to make it accessible? And you're right, people have been working, we've been working with companies for several years now to try to bring things up and it's based on priorities. But what are the things you're seeing that they're going to need to do to be ready for June, June 2025, eight months from now?

Bill Kasdorf

There's a whole bunch of easy things, and there's also a bunch of hard things. The easy things are things like every HTML file needs to have a language tag in the HTML element at the top, right? That's just really simple, but 80 to 90% of the books I review don't have that. They specify the language, but they don't use the proper tag. And so that's easy to just get, right, right? You need to have proper structural markup, which means your headings have to be in the right order without – nested properly with no skipped levels. That's really important, because that's how a person with assistive technology understands the structure of the book. So they don't want to have to just read through it until they find something that they're looking for. They should be able to drill down by sections and say “Oh, that's what this section is. I see what the heading is. I don't need to read all these subsections. I can skip to the next page too.”

They also need to have a link table of contents, and it should be expressed in what's called the toc nav. A nav is an element in the HTML because, and many books do have that. One thing that is not a requirement, but I see a lot in books. I've got a couple big accessibility reviews I'm in the middle of right now, and ironically the EPUBs have links at the top of beginning of where all the page breaks are. And the reason is that the index locators then point to those tags, which is just the same. All the index is telling you is it's somewhere on page 52. So they've got an A tag for page 52. But the problem is, that doesn't work for what's called a page list in EPUB that enables navigation, pagination-based navigation, which is also really useful. But they've already found all the spots.

36:00

They've already marked them all, so it's just changing the syntax that you've got at those places and that could be automated. So I consider that something people don't think of. And then you have a what's called a page list nav in addition to your toc nav. So that enables another way of navigating to a disabled user.

Mark Gross

And those are all, like I say, they're easy, they're global, and they're really navigational. And I think it'd be useful. I mean, if you watch somebody who is visually disabled working on an eBook, you immediately get to see what are the important things, that might be a – that might actually be its own webinar at some point.

Bill Kasdorf

I said, those were the easy ones off the top of my head, let me mention a couple of hard ones. Let me mention a couple of hard ones. Here's one that is technically easy but hard to implement, and that is, okay, you've got your lang equals en or en-US or en-GB tag at the top of HML. That's called the primary language. But you also should be tagging any words or phrases that are in a different language, those should get tagged. And the reason for that is, it's not a translation issue. I've got a book right now that I'm analyzing for a client that every time they've got a foreign word or phrase like that, the English equivalent is right there. But it's about pronunciation, because the screen reader technologies, these are really amazing technologies, by the way. It's really cool what they can do. So, I had a book last year sometime, I think it was, from a publisher, that it was either about the use of Hebrew in Arabic literature or the use of Arabic in Hebrew literature. It was one or the other of those, I can't remember which one it was, but holy moly, those aren't just not English, they're not Latin script. And they read right to left. Both of them read right to left. So this is not trivial, right?

Mark Gross

They are Unicode.

Bill Kasdorf

That's the solution. So, one of them had images of all these things, but my pointed out to them that their typesetter had to create that image. They probably typeset it with a Unicode font. They probably actually have the Unicode somewhere that you should be putting in your EPUB. But another one of these books like this, that is what, they were actually text. They weren't images, which is good, but they were just Hebrew text and Arabic text. I happen to have a friend in the business that is an accessibility expert, and he happens to speak fluent Hebrew and fluent Arabic. So I sent that book off to him and said “Let me know what you think about this,” right? And he said “Well, when I first started using “I think it was using NVDA, which is –NVDA and JAWS are the two most prominent and most highly evolved screen readers used most by most people. And it was like gibberish. But he said “All I had to do was load, I think it was called the Hebrew vocals and the Arabic vocals to that to my screen reader.” And he said “I could read this entire book and it just read me the Hebrew properly and it read me the Arabic properly.” In Hebrew, we're not talking about translation, we're talking about –

40:00

because the book is designed for somebody that can read Hebrew and Arabic. That's all that's there. There isn't any English in this thing, but it really matters. Now, I had a client, one of the things I often do is subcontract to an actual screen reader tester for books or websites, because I want to get their lived experience of what's working here and where are you bumping into things that are serious problems for you when you're doing this. But they tend to be pretty doctrinaire about their responses. So one of the things that they said needed to be done was that every place in this, and it was a scholarly book, every place in this book that said “et al.” needed to be tagged as well. Well, do you think I recommended to my client that they have to fix that book and tag those all as Latin?

No, that's ridiculous. Any reader of that book knows what that is, and a screen reader is going to pronounce it fine. you don't have to go back in and fix that. Let's get real here. But a lot of this is judgment-based. There are a lot of rules. Some of these things, once you get, they can become routine and particularly by your vendors, et cetera. But one reason that I said this language tagging is harder is just finding all those places in the book that are in those non-English, that are non-English words or phrases is a challenge. I used to have a typesetting business and people thought that the typesetter read the book. No, the typesetter doesn't read the book, they're just typesetting the book.

So it's really upstream. The editors and the copy editors are the ones that, and maybe the proofreaders are the ones that are actually reading every word in the book. So, that's probably a workflow issue that might be a little bit of a challenge to get those people to mark those places so that your vendor can know where to put the tags. Because if you're paying them to find the tags, that's going to be expensive. And then the last one is what everybody's flipping out about, and that's image descriptions, because, that's another conversation, but image descriptions are the big one, and they are very important.

Mark Gross

Alt text. That's a place I think where people are, those are things that are typically have been left out and need to be done here. What's also related to that is table structures and math, right? A lot of those are set up as images, if you want math out there, if it's an image it's very difficult to describe. Whereas if it's done as MathML or one of those other coding sequences, the screen reader can actually read those. And tables are the same thing.

Bill Kasdorf

Exactly right.

Mark Gross

What have always been to put images in. And when you're dealing with this it's not good enough.

Bill Kasdorf

You might still want an image in your EPUB, but under the hood you still got to have MathML or you have to have an HTML table. And the thing about tables is that, well, first of all, in a lot of vendor workflows, I used to work for one of the main vendors and one of the – when we took over doing PLOS and just dramatically decreased the turnaround time for them to get finished publishable articles, the key was we converted everything into detailed JATs with MathML as soon as we got the manuscript. And then everything else could just flow from that. It was great. But, so, lots of your vendors know how to do MathML. And the thing about tables, they might have been HTML tables upstream in the workflow too,

44:01

because somebody had to typeset that table to generate the image, right?

Mark Gross

There's a distinction –

Bill Kasdorf

You need to mark –

Mark Gross

Your ongoing workflow. And my focus a lot of times is on the backlist and what do you do?

Bill Kasdorf

That's right.

Mark Gross

A million pages you did over the last few years.

Bill Kasdorf

That's why the backlist is such a challenge because those are already done and now you got to go back and fix them.

Mark Gross

Right, so those are the challenge and those, I think that's where people are going to – which speaks to, I think to the next question is a term that's used several times over here, disproportionate burden.

Bill Kasdorf

Yes, that's an important one.

Mark Gross

And that seems like a major loophole over there. How do you describe disproportionate loophole? I mean, to one person anything is disproportionate, to another person it's “No, well, you got to do it anyway.” What's the discussion out there on that?

Bill Kasdorf

Well, I would prefer not to focus on the discussion, which is often ill-informed, and point out that a lot of people think that this directive is impenetrable and hard to understand and doesn't explain all this stuff. They just make you do things and they don't explain it, right? There's actually very clear instructions about how they judge whether there's a disproportionate burden. It has to do with a handful of different factors, how big the organization is, what the cost involved is, what the time involved is to deal with this, what kind of sales could be expected from this thing once you get it fixed, et cetera. So mostly what you need to know about, one of the things is micro-enterprises are exempt. But one of the things you need to know is – and this is why I say you really need to have your counsel, or your legal counsel analyze some of these things – is it actually does pretty much spell out, I'm not going to get into the details now, but it does spell out how to tell if something qualifies for disproportionate burden. And that's not a publisher as a whole, it's an individual product. I've got clients that have –

Mark Gross

Guidelines –

Bill Kasdorf

twenty-year-old backlist books that would be a nightmare to update, but they sell five copies a year. Well, I'm pretty sure that's going to be qualified for disproportionate burden, because it's going to cost them thousands of dollars to do that and it's, they're only going to make a few hundred bucks off it.

Mark Gross

Right, right. I think I did read, I was looking, it seems more like guidelines than actually –

Bill Kasdorf

Yeah, that's right. But it does tell you, here's how we determine whether there's disproportionate burden or not. It is in there, but it's not all in the same place. That's one of the things why it's kind of difficult, because it's using terms that are now defined in the back, in the definition section.

Mark Gross

Right.

Bill Kasdorf

So just reading –

Mark Gross

It comes back to what is good business practice also. I mean, I don't think it's – “Get the lawyers in” always scares me a little bit, but I think it's also a good practice concept over here. If you make your materials accessible it's better business, people, your books are the same, they're workable the same, and it should be good business. But like you say, the book that has, that you can sell five copies of, that's probably not going to be worth the whole bunch to be done. You'll just get a reader to do it and read it to them, right?

Bill Kasdorf

That's right.

Mark Gross

Oh, Marianne's here with questions.

Bill Kasdorf

It looks like we're running out of time. Here's Marianne.

Marianne Calilhanna

We do have quite some questions.

48:00

We actually had a number of questions about disproportionate burden, so I'm glad you did discuss that. And I do invite, if we don't cover your question, please feel free to reach out to Bill or DCL and we'll absolutely follow up after the webinar. One person asked if you could just clarify, Bill, what version of WCAG are we on? Is it 2.0, 2.1 or 2.2?

Bill Kasdorf

It continues to evolve. The most recent is WCAG 2.2.

Mark Gross

It is-

Bill Kasdorf

That is not required. A, it's too recent. It's actually not that much different from 2.1. There are aspects where 2.0 is sufficient, but the thing is that pretty much anything that's in one version is augmented not eliminated from the subsequent version. So if that thing is in 2.0, it's probably in 2.1 as well. This is not going to be true when WCAG 3.0, that's a whole different animal, that's going to be completely different. I don't want to panic people. But yeah, it's pretty much going to be completely different, but it's years off. So, the general advice that I give, and I think most people discussing this give is use WCAG 2.1 AA as your guide.

Marianne Calilhanna

2.1 AA, okay.

Bill Kasdorf

Yeah, and A, we know that EPUB accessibility 1.1, if you comply with what it says to do, you'll be conforming to WCAG 2.1 AA. There are nuances because there are certain things that are musts and certain things that are shoulds. So, I don't mean to imply that it's all cut and dried, it's not all cut and dried, but that's the general guidance. And one thing that we haven't mentioned is that there was a recent change just came out in April in Title two of the Americans with Disabilities Act. Now, I don't know if you want me to get off onto something that's a little tangential here because it's not the EAA. The point is it goes into effect, let me see here. April 24th, 2026 for large organizations, which is over 50,000 people. And April 24th, 2027 by smaller organizations. It says public entities. And in the kind of explanatory documents from the government on this they use the term state and local governments, instead of saying public entities, because that's ambiguous, people don't understand clearly what that means, but it isn't just the governments, I mean it includes public schools, public universities, libraries, et cetera. So, if you think “Oh, well, we don't sell to the government,” well, you may want to step back and think “Oh yes, we do because we sell to libraries that are public libraries that are subject to this Title II of ADA.” And that requirement is also WCAG 2.1 AA. So it's like just do it, right? Just do it.

Marianne Calilhanna

Okay. We've got a number of questions coming in fast and furious. So one person wrote “I've heard the image descriptions will not be required by June for STM journals. Is this true?

51:59

And are you aware of any AI tools that can generate accurate alt text for biomedical images?”

Bill Kasdorf

I'll take those in reverse order, but I'll try to go fast, because we're running out of time. The answer to the last one is no, progress is being made. So I think certainly for doing a large volume of images, AI, well implemented AI can be a valuable tool, but you always need human review. One reason this is important and people forget or don't realize that there's two different kinds of image descriptions. There's alt text, which is just plain text and it's very short. Usually people use guidance like it's like a tweet, or it's like 250 characters or so. And then there's what used to be referred to as long description. Sometimes it's still called that. But in the HTML world it's called extended description. And the distinction is that that can be of any length, but also it can include markup. For example, there's an image description expert that I've known for years that, well, I'll mention who it is. It's Huw Alexander from textBOX in the UK.

One of his clients is Cambridge University Press. And they had a, I don't know what it's called because it's linguist, but it's like a flowchart diagram in a linguistics journal. And apparently these things come up all the time. Well, to you and me it's a bunch of lines, some intersecting, some not. And where at the nodes there are letters and numbers on them. It's like that's not an adequate description. It does describe what that thing looks like, but this is a linguistics journal. So what he did was he involved linguists and came up with just this brilliant extended description. But yes, it's highly structured. It's got sections and it's got, because there are different steps that are involved here, and then it's got lists in there, et cetera. So for that particular image, for the audience that it has attended for, that's a good extended description, but that's what's needed.

The principle is the image description, and we try to emphasize using that term, don't just call everything alt text, because that's a specific thing. But the image description should convey to a non-sighted user what the image conveys to a sighted user. And this means that, and it should also not repeat what's in the caption or in the surrounding text. So I saw an example from a different publisher, a different context, but they had all of these kind of interesting, almost hieroglyphic diagrams and things, and there were a bunch of them in this book. And so, I was sent the images and in the context of the book, and they said “These are in fact completely described in text where the image occurs in the book. What are we supposed to do? Because we still have to have alt text. It's a requirement.”

Image, extended descriptions are not a requirement. Alt text is absolutely a requirement. It's an HTML requirement. And I said it's simple. Just say “Image is fully described in the adjacent text.” That's your alt text. That's fine. So it's not an empty alt or a null alt because that's illegal, but that's all you need to say because it's all described right there.

56:00

But that's why this is so complicated because AI can't pick up on those nuances very well. And particularly the context issue and audience issue –

Mark Gross

Simple things on images themselves. It's not going to really do a particularly good job, other than describing there's a person in the field. I mean, it's really not, they're doing a great job of that. But when you get into those specialty languages in linguistics is one of them that quickly can go into that disproportionate of burden area, I imagine. So I think that needs to be looked at. These could be hundreds of dollars per image.

Bill Kasdorf

This is also why I've advocated for years, way before there was an EAA that for scholarly publishers and the more, well, let's just leave it at that because we're running out of time. But for scholarly publishers, I've always been a big advocate of the author who, if the author is providing the images, they should provide what I call a draft image description. They won't necessarily get it right, but they know why that image is there. And if it's a linguist, they know how to describe that image, because you would describe it to one of their linguist colleagues, and they should be the one providing you that. And then your editors should know what makes a good image description. And that's a fundamental job of a publisher, is editing. So those are text. So if you're going to edit the book, you should be editing those descriptions. And if you use AI, consider those the same thing, draft image descriptions. Okay, this is getting me a good start, but now I need to tweak it to make it actually right.

Marianne Calilhanna

Right. Okay, we're not going to get to all the questions, so we will follow up with individuals. And I think this is a good closer: “Looking ahead, Bill, do you think the EAA will shape the future of digital accessibility in Europe and potentially influence globally?”

Bill Kasdorf

Absolutely. It's going to have a huge influence, partly because the market is global. And many of these things, as Mark pointed out, are very difficult to do with the backlist, because there can be, I mean, like I said, from a trade publisher like PRH has thousands, but a scholarly publishers like Taylor & Francis also doing a wonderful job on this stuff, thousands and thousands and thousands of books in their backlist that are not accessible, but they're dealing with it. They're all working systematically to do what they can. But for frontlist, a lot of these things are just going to become routine. It's like this is just how you do a book. So you won't get books with messed up markup and you'll get metadata with your book, et cetera.

And so it'll become more taken for granted going further. But it's frankly getting a gun to our heads with the EAA that's making people sit up and say, because everybody knows that they should have been doing this, right? Nobody argues that you shouldn't have your books be accessible. But it's like those of us of a certain age remember when SGML was called “Sounds Good; Maybe Later.” For years and years, that's what the attitude toward accessibility was, right? Yeah. I've been advocating accessibility for over 20 years and nobody ever disagrees with it, but nobody was doing anything about it. Well, guess what? Now they're doing something about it.

Marianne Calilhanna

Right.

Bill Kasdorf

So thank you, EAA.

Mark Gross

Technology is available, and the tools are available, and now there's some legal support for it. But I mean, already for a number years, I mean, you can't sell a book in the state of Texas if it doesn't have certain standards on. So this business reasons for doing these things also, that's ultimately driving a lot of this, I think, going forward.

Marianne Calilhanna

Yeah.

Bill Kasdorf

Absolutely. One of my clients a year or two ago was a K-12 publisher that was going after, a very small publisher, but a very good one that was going after a Texas adoption. But they were working crazy to get their books or their resources, including things like worksheets and stuff, as accessible as possible. And that's why they came to me is to help them do that and to give them guidance on how to do it in an ongoing way. And I was just talking to my contact there a month or so ago, said “By the way, did you ever get that Texas adoption?” She said “Yeah, we got the adoption.” So it was worth the work. But if they hadn't done that, they wouldn't have gotten the adoption –

Mark Gross

The business reasons for it.

Marianne Calilhanna

Yeah. Well, we have come to the top of the hour.

Mark Gross

Their books were perfect. They were fruitful.

Marianne Calilhanna

Thank you both so much. We have come to the top of the hour, and I want to thank everyone for attending this webinar. The DCL Learning Series comprises webinars such as this. We also have a monthly newsletter and our blog. You can access many other webinars that are related to content structure, accessibility standards, and more from the on-demand section of our website at dataconversionlaboratory.com. We do hope to see you in future webinars, and this concludes today's broadcast. Thank you very much.

Mark Gross

Okay. Okay. Thank you. Thank you, Bill. And thank you everybody.

Bill Kasdorf

My pleasure. As you could tell, I like talking about this stuff. Thanks for giving me the forum.