1 of 27

What’s the Deal with PDFs?

Sagan Wallace

OSU Libraries & Press

2 of 27

Today’s Agenda

How PDFs work

PDF accessibility

Decision-making strategies

1

2

3

So what are we covering today? This is going to be a little bit of a meandering conversation. I’m working towards a conclusion and I have specific things I want you to learn as we go, but unlike how I normally structure my presentations with a clear set of learning objectives, this one is a little more philosophical. I’ve been jokingly referring to it as my Manifesto, if that helps. Or scares you away.

In general though, we’ll be going over these things. I’m going to talk about how PDFs work, beyond “you open them and you read them.” I’m going to talk about how they can and cannot be made accessible, and some strategies, alternatives and problem-solving practices you can consider. And I am hoping to leave with lots of room for Q&A but if you have follow up questions I’m always open to meet and email outside of this.

So to start, I’m going to see who falls on which side of the argument.

3 of 27

Team PDF

PDFs are…

How scholarly communication happens
How research is shared
Protects my work
Maintains document integrity
Universally compatible

Some of you are strongly on Team PDF. You have really good reasons. PDFs are how scholarly communication happens - particularly in the sciences. Research is shared via PDFs, from start to finish you are generating and distributing PDFs.

It’s I Can Haz PDF, not I Can Haz EPUB.

Not only that, but PDFs are static. It protects your intellectual property, and it means we can discuss and cite concepts without worrying that they are going to change afterwards. It’s a universally compatible format - you can be on Windows, Mac, or your jailbreaked device and you can still read the PDF and see it the same way I see it.

Image credits:

https://en.wikipedia.org/wiki/Adobe_Acrobat#/media/File:Adobe_Acrobat_DC_logo_2020.svg

https://www.pexels.com/photo/woman-sitting-on-the-floor-among-laptops-and-tangled-cables-and-wearing-goggles-8721343/

4 of 27

Team never PDF

Hello Darkness, My Old Friend…

PDFs are…

Impossible to make accessible
Can’t be updated
Can’t handle math
Horrible on mobile
Should die already

5 of 27

Can’t we all just get along?

It really makes you ask, why can’t we all just get along?

Both sides have important points to make.

Both sides are willing to fight it out.

I’m part of a task force on campus to help decide how to handle our STEM-heavy incoming electronic theses and dissertations. It was called by the Grad School and they asked for both STEM faculty and accessibility faculty to attend to give input. And when I tell you we were literally divided on one side of the room from each other, trying to battle out for whether or not we would collect our ETDs as PDFs or a different format. People get heated.

I tend to be on the anti-PDF side, out of a lot of experience of trying to remediate PDFs. But I don’t want to spend today dismissing PDFs entirely. Instead, my goal for today is to give you the information - as bias-free as I’m able to do, about how PDFs work, what it takes to make them accessible, and how you might navigate this as you make decisions about your online content.

Image credit: disneyplus.com/en-gb/browse/entity-2bd19267-1387-4b51-98e6-5189ff28e9c9

6 of 27

01

What is a PDF?

7 of 27

What are PDFs?

“Portable Document Format”
Invented 1993 (Project Camelot 1991)
Multiple open standards: ISO 32000

PDF/A
PDF/UA

Contains code to generate layout and graphics, embedded fonts, and internal storage

John Warnock

(1940-2023)

PDF stands for Portable Document Format. It was created by John Warnock, pictured here, thank you Wikipedia, as part of Project Camelot, in the early 90s.

It uses a set of open standards, specifically ISO 32000, that allow you to easily set up your documents precisely and in a way that can be easily reproduced. For example, PDF/A is a standard version that makes a file suitable for archiving, and PDF/UA is a standard that states the document is digitally accessible.

It’s a remarkable content type because it encodes layout, graphics and text in a way that means no matter who opens it and wherever they open it, they see the same materials.

I can’t emphasize enough how important this is. Imagine you’re signing a legal document - you want to make sure every detail is exact, right? Or, have you ever been on a hiring committee and gotten a kind of weird looking resume, and tried to figure out if that’s because the formatting got messed up between their mac and your PC, or if it’s evidence that they’re not a detail-oriented person. Or, are you any of the universities that use GradeScope to make grading your large classes manageable, which functions by perfectly aligning the answer squares for the homework you assign? PDFs are more than a pretty face, they made it possible for us to digitize our world.

How does it do all this?

Image credit: https://en.wikipedia.org/wiki/John_Warnock

8 of 27

Parts of a PDF

Visual

Content

Tags

I really like this visual from WebAIM about what a PDF is “made” of. It’s easiest to think about it as having several different layers.

The layer on top is the “Visual” layer. It’s what you see on the screen when everything is rendered and working together. It holds the content in place so it perfectly matches the “source” document, no matter what device or operating system you’re using.

Beneath that is the content layer. It’s the same information, with all but the basic formatting removed. So in thai case, you can see the same lines of text and the image is still there, but it’s no longer centered.

Underlying all of this is the Tags layer. This is the actual structure of the document, and it’s this layer that makes a screen reader fully work. This is the language that connects the machine that’s holding the PDF - your computer or laptop - with the machine that reads aloud the document (your screen reader).

Image credit: https://webaim.org/techniques/acrobat/#important

9 of 27

Visual

Parts of a PDF

Content

Tags

10 of 27

Content

Parts of a PDF

Visual

Tags

11 of 27

Tags

Parts of a PDF

Visual

Content

12 of 27

Examples of Tags

13 of 27

Creating an Accessible PDF

Hard way:

scans from print,

OR

created in Google Docs, Apple Pages…

Easy way:

Source document Microsoft Word, PowerPoint, Overleaf, Canva (sorta)…

When we’re talking about PDF creation, we’re not talking about starting in PDF format. PDF is converted from another format. And there are easy ways and hard ways to do that accessibly.

Some methods are always going to be hard. Like we talked about scanners earlier, digitizing print material will always be a remediation step by definition because you are starting with a photograph. You can also do it the hard way if you are using document creators that can’t export tagged PDFs - Google Docs and Apple Pages, or old versions of Word, are examples of this. If you create your materials here, you will force yourself to have to remediate your PDF later.

Other platforms can get you part of the way there. Canva is my go-to example. Canva used to be very inaccessible but they’ve made moves to fix their product, so you can now export a tagged PDF. However, last I tested, it doesn’t reliably export tables, and it places character limits on alt text so you may need to fix that later.

The easy way is when your document is born digital. In these instances, you have access to a source document. That source document can be set up so it exports a tagged PDF. Sometimes this is relatively straightforward. Microsoft Word is a good example. It’s very well integrated and creates a PDF that needs very minimal post-processing - usually all I need is to add a title, and even that you can do in Word first.

14 of 27

Creating an Accessible PDF 2

Hard way:

scans from print,

OR

created in Google Docs, Apple Pages…

Easy way:

Source document Microsoft Word, PowerPoint, Overleaf, Canva (sorta)…

15 of 27

When is accessible not Accessible?

Physical spaces follow the ADA Standards for Accessible Design (1991, 2010), which includes very specific requirements

Digital spaces follow Web Content Accessibility Guidelines 2.1 AA

OK so you’re using Microsoft Word, you’re following all the steps to make a tagged PDF, and you’re still hearing that “PDFs should die already.” I thought we’ve created an accessible PDF, right? Why isn’t it good enough?

There’s a difference between generally accessible, and what is required by law

such as:

Width requirements for hallways

Angles of ramps/curb cuts

Ratios of accessible parking spaces

Etc.

We don’t wait to build an elevator until a wheelchair user requests one; we embed it into our construction.

, which include specific requirements such as:

Content is read in a meaningful sequence

Contrast ratios of 4.5:1

Content can be resized to 200%

Content reflows to 320 CSS pixels vertically, 256 CSS pixels horizontally

Etc.

So what does it take to make a PDF accessible to WCAG guidelines? There are four main hang ups that can come up at different points - alternative text, math, reflow and PDF content issues.

Image credit: https://www.access-board.gov/ada/guides/chapter-4-ramps-and-curb-ramps/

16 of 27

How Alt Text Works

If alternative text is present, it is announced by the screen reader:

It is read without audible pauses
It cannot be paused once it begins to play
It is not rich text (no foreign languages, no symbols

(1.1: Provide text alternatives for any non-text content so that it can be changed into other forms people need, such as large print, braille, speech, symbols or simpler language.)

17 of 27

Example

Multiple methods of providing full alt text:

Use the alt text field
Describe data in the prose of your text
Provide a data table

18 of 27

Let’s talk about math

PDFs usually put math equations in a <figure> tag, using alt text to provide meaning:

“The square root of A plus B over C”

19 of 27

Let’s talk about math cont’d.

“The square root of A plus B over C” is ambiguous.

20 of 27

Alt text is plain text

′＇´ ՛

an apostrophe
a prime
an acute accent
the Armenian accent symbol

A^O A^໐ A°

A to the capital letter O
A to the capital zero
A degrees

Unless symbols are coded correctly, typography that can be understood through context visually can be misinterpreted by a screen reader.

What do the following symbols mean?

21 of 27

How does math become accessible?

MathML + MathJAX = perceivable and operable math

<mtr>

<mtd>

<msub>

<msup>

</msup>

</msup>

PLUS A LOT MORE CODE IN THE MIDDLE

</math>

OMML→MathML→PDF→<formula>

With all the caveats that there are always multiple ways to solve a problem, there is “industry standard” for how to handle math: it’s written in MathML, then the document is read using a platform that can incorporate mathJAX (usually a web browser), so it can then be announced meaningfully to a screen reader.

MathML: the language math is written in that most screen readers can use. Here’s my complex math equation, and beneath it I have a very brief snippet of mathML code. This mathML is understood by

MathJAX: a Javascript display engine used ubiquitously by screen readers and web browsers. An equation written in mathML can be navigated using only a screen reader and their keyboard. It can be converted to Braille. It can announce special characters.

(n.b.) You might see references to OMML: Office MathML, which allows MathML to be integrated into Office applications (Word, etc.). When you use the equation editor in Word, it is writing it in OMML. That OMML, when the word doc is converted to HTML, EPUB, or other compliant formats, gets automatically converted to mathML. I personally have have a 100% success rate of converting math in Word to HTML/EPUB, but older versions of Word might cause issues, and you’ll see some info on the web saying that OMML is not accessible. To my knowledge Microsoft has made a lot of improvements and this is no longer true.

When you convert from Word to PDF, content written in OMML gets converted to a PDF <formula> tag. This takes the background mathML, creates an alt text based on how it would have been read aloud from html, and adds that to a <formula> tag.

https://www.w3.org/Math/whatIsMathML.html

https://www.mathjax.org/

22 of 27

LaTeX

Typesetting program heavily used in STEM
Under active development to allow PDFs to handle mathML

23 of 27

Reflow

WCAG 1.4.10 (AA)

Content can be presented without loss of information or functionality, and without requiring scrolling in two dimensions for [specific pixel widths]

24 of 27

My Misc. Gripes

Some PDF errors cannot be fixed without returning to the source document

Incorrectly embedded fonts
Error message: “MCID <x> already present.”
XRef table corruption
Color contrast issues

25 of 27

How do you make a decision?

PDF accessibility guidelines are not the only need

Archival needs (bit rot)
External requirements (ex: grant funding)
Funding
Intellectual property

Remove PDFs where possible so you can focus where PDFs are necessary
If you can help it, never remediate a PDF, focus on PDFs you have control over
Aim for flexible formats
Think outside the box

26 of 27

Final note: PDF forms

27 of 27

Thank you!

sagan.wallace@oregonstate.edu