Keep It Silly Simple
I sat down to review some code with my colleagues. It was clear that each of these solutions was heavily over-engineered. Each used custom web components, React, Angular, or other framework, and even basic HTML with aria added. All of them should get slimmed down and work to reduce or remove ARIA.
Comments closedWe previously discussed contrast under Guideline 1.4.3 Contrast. So why is there a second AA criteria for contrast? The earlier guideline covers only text. With the introduction WCAG 2.1, we need to start being aware of the contrast of graphical items, particularly active controls. But the way we test contrast for graphics is different than how we test contrast for text, kind of…
Just as we did in testing text contrast, we’ll need a contrast checker. There are many out there, but for this article I’ll be using Level Access’s Accessible Color Picker. The one you use may be different, but the concepts should be the same.
The visual presentation of the following have a contrast ratio of at least 3:1 against adjacent color(s):
- User Interface Components: Visual information required to identify user interface components and states, except for inactive components or where the appearance of the component is determined by the user agent and not modified by the author;
- Graphical Objects: Parts of graphics required to understand the content, except when a particular presentation of graphics is essential to the information being conveyed.
Technical words for technical people. Welcome to the world of writing standards!
“The visual presentation…contrast ratio of … 3:1 against adjacent colors.” Cool. We’re familiar with 3:1, we know how to use the contrast checker, let’s test!
There is a key phrase here that complicates the testing, “adjacent colors.” Our control must have a 3:1 contrast ratio against adjacent colors. Colors. How often have you tried to make a color palette with three colors that all have a 3:1 ratio? If you haven’t yet, it can be quite difficult. The more colors you add to the palette, the harder it gets.
But this adjacent colors is not expecting you to have a 3:1 contrast between all colors. The second bullet contains important information too. “Parts of graphics required to understand the content…” Of course there is more to it, but let’s start here. We’ll start with a simple button for a disclosure. Let’s get our example from WCAG.
This button has blue text on a gray button. The button has a darker gray border and sits on a white background. Most people will test the background color of the button to the white. And that is one way to do it. In this case, it will result in a false positive. First, because there is a border, the gray background is not adjacent to the white. So we measured the wrong thing. We need to compare the border with both the white background and the gray button background. This is because the border defines the edge of the control. This success criterion is about discerning actionable controls.
When we do this test, we find out that neither passes.
The border is #E1E1E1 and the gray only has a 1.127:1 contrast.
When the border is against white, it only bumps up to 1.308:1. But we’re not done testing this unit. We’ve checked the border against adjacent colors and it fails. But WCAG gave us an out that the W3C is using on their own site. “Parts of graphics required to understand the content…”. What is important to understand the content? Is it the background, the border, or is it something else, say the text of the button.
The text on its own has enough contrast against the gray. No additional graphic is needed to understand the control.
As seen in this image, the blue text against the gray has enough contrast to meet the 1.4.3 Text Contrast rules, and it still supports 1.4.11 despite the background failing.
When testing 1.4.11 non-text contrast:
Thoughts, questions, have a different method? Reach me on LinkedIn and BlueSky to discuss the topic!
1 CommentWe live in a time where our digital device screen sizes vary. We have 1 inch watch screens, phones, tablets, and even TVs with 100 inch screens. We never know what the end use will be using for a device. Reflow is needed to adapt to many of these screens. However, the core of what reflow is about has little to do with screen size. Let’s look at what we really need to know about 1.4.10 Reflow.
Looking at the WCAG Quickref, we see the following:
Content can be presented without loss of information or functionality, and without requiring scrolling in two dimensions for:
- Vertical scrolling content at a width equivalent to 320 CSS pixels;
- Horizontal scrolling content at a height equivalent to 256 CSS pixels;
Except for parts of the content which require two-dimensional layout for usage or meaning.
Reflow talks about having no loss of functionality without having to scroll in two directions on screens as small as 256 pixels wide. It tells us that when the screen is at 256 pixels wide we can have horizontal scrolling, but not vertical. At 320, we can have vertical but not horizontal scrolling.
As a matter of good web principle, it is better to design vertically than horizontally in the majority of cases. For news, reading, and shopping it is easier to consume the content and it is the expected method of scrolling.
Remember, we are talking about web content accessibility guidelines. For whatever reason, every time I discuss reflow with developers they automatically think it is for mobile devices. And yes, it benefits everyone on any device when implemented correctly. But so does just about everything listed under WCAG (Not all screen reader techniques benefit the able-bodied). And there is a key word that gets left out during those conversations.
This guideline is about making websites more accessible. Reflow wants to ensure that the site does not hide any functionality when any user is on the site. It should not obstruct, obscure, or remove functionality. This includes when that user adjusts their monitor to a different resolution.
A huge number of people on this planet have issues with vision. For most of them, a pair of glasses will correct it. We’re told by our doctors that 20/20 is the most accurate a human can expect. When a person’s vision reaches 20/200 after correction, they are considered legally blind. They have some vision though and decide to increase the resolution on their screen to read it easier. Likely they may not be aware of screen readers yet. Our volunteer here increases his screen resolution to 300% at which he can read it.
When discussing this addition to the guidelines, 1280×1024 was the most dominant screen size for desktops and laptops. There was also thinking that most disabled people are on fixed income and can’t afford current tech. This lead to thinking the disabled to be on older devices.
If we were to increase our zoom from 100% to 400% starting at 1280×1024, the page is rendered at 320×256. If something requires two way scrolling at this point, you need to revisit the reflow rules. We do not want to remove functionality, but it is ok if functionality goes into a menu or disclosure. We want to stack content in the scroll direction.
There are pieces of content that can have horizontal scrolling including carousels and tables. With carousels, ensure the controls are on the screen, but the contents can overflow. You can swipe, but provide buttons for fallback. For tables, ensure the header column is always visible with at least one other full column.
I’m not writing to a generic audience here. I hope you have deduced a method for testing, but I won’t leave you hanging. Testing is simple, set you browser viewport to 1280 by 1024. Load the page, then use Control/Command and the plus (+) key until you reach 400% zoom. Review the content to ensure nothing is removed. Make sure content is stacked or hidden by disclosures. Ensure there is no horizontal scrolling. You can flip the dimensions for horizontal scrolling sites.
Alternatively, open your developer tools in mobile emulation. You can leave zoom at 100%. Change the viewport to 320×256.
Have some thoughts or more questions on Reflow? Find me on LinkedIn or Bluesky to keep the conversation going.
3 CommentsI’ve been in developing websites for over twenty years. When I started in the industry, we had to create images of text if we wanted to use a specific font. This was necessary to make things work the way we wanted visually.
The problem of images of text is that screen readers can’t see the text. Like other images we need to provide alt text that reflects the text in the image. It was early in my career, so I tried to find a way to write. I used individual PNGs of letters with the alt text set to that letter. What I didn’t fully understand was assistive technology. My attempted method ended up with the AT announcing each letter. This is because they read each image as a separate unit of text. Spelling my name would be announced as, “N image, A image, T image.” This doesn’t help anyone.
The folks at the W3C’s CSS working group have been working overtime. They have provided us ways to import fonts from CDNs. We have the ability to clip with text. Drop shadows, gradients, transforms provide additional customization.
We also have SVGs. They allow us to put text directly into an image as text. This text is available to AT. This allows us to put stylized text on a path for display.
We do not need to put text in our images anymore. And I’m not referring to logos. That another can of worms. We have multiple ways to add stylized text with unique fonts. We don’t need to use text in images anymore. Just don’t do it.
I know what you are thinking, “My current platform doesn’t allow me to…” or “That requires me to work with IT to make changes and I don’t have many hours…”
There are always constraints. What I described earlier is a best case scenario where all conditions are right to make this problem go away. But that’s not how life works. You will need to use a custom font as an image of text at some point.
And it isn’t difficult. We treat it just like we do with other images. We use alternative text to supplement the image with a text equivalent. In the case of text in an image, that includes all the text.
If you alt text is over 150 characters, I want you to pause. When alt text gets really long it can overwhelm some users. Neuro-typical brains will probably be fine. But neuro-divergent users will have problems. This may be an area where you can break up the image into smaller images. Break the text up to match the images. You can also use figure and figcaption. Here, everything is wrapped in a figure. The alt text is placed on the image. The figcaption contains the rest of the explanatory text.
<figure> <img src="https://placekitten.com/200/300" alt="Text in the image." /> <figcaption>The remainder of the the text described in the image to complete the message needed to express things.</figcaption> </figure>
And the most important tip I can leave you with is:
Always put sentence ending punctuation at the end of your alt text.
When you don’t use closing punctuation, most AT browser combos will start reading the next item. They will read it as if it were a run-on sentence. This causes added confusion. With the punctuation in place, the AT will pause or announce the punctuation. This is determined by the user’s settings. In either case, the user understands the concepts better without the run-on sentence.
Have some thoughts on this topic? Let’s talk over on LinkedIn and BlueSky!
2 CommentsUnless you have bionic eyes, at one point or another a computer screen will become difficult to read. For most of us, we’ll reach for glasses or call to make an appointment for an eye exam. But at some point, the glasses just aren’t enough.
I just got my first pair of bifocals. I’ve been using two pairs of glasses – one for distance and one for reading. Before I got the bifocals I had already increased the resolution of my monitor to 125%. All of this to put less strain on my eyes while reading.
And that is the core reason for 1.4.4 Resize Text.
Following WCAG’s QuickRef, we see:
Except for captions and images of text, text can be resized without assistive technology up to 200 percent without loss of content or functionality.
Without assistive technology our digital assets need to increase their font size to 200%. PDFs you can zoom in on to increase the font size. To test in a browser, press CTRL or CMD and the Plus key. The browser will show you the zoom level as you do this.
Now with the page zoomed to 200%, we need to look for the following:
You should have no problems if you’ve ensured that your fonts and containers use relative sizing like EM, REM, or percentage. This also includes viewport width or height. Text and containers will resize smoothly.
It’s 2025. Not all applications are feasible on mobile, but we need to rid ourselves of this thinking. It’s 2025 and there are so many screens in the world, we will never know what the user has. We must make our interfaces flexible. Responsive design is not “feature” we add on to be nice. It is a requirement.
Users with low vision don’t start with third-party assistive tech. They start with what they can adjust on their device. When they zoom in from 1280×1024 to 200%, the view should match what is seen on a mobile device. Comparatively, this device would have a 640×512 pixel resolution.
Enjoy this post? Have questions on it? Find me on LinkedIn and BlueSky!
2 CommentsOne of the most common issues I see when testing is the content not having enough contrast. With 1.4.3, we’re focus on the contrast of text. This can be text in an image or just on the page. Later we’ll talk about the contrast of active controls in 1.4.11.
Designers commonly complain that they can’t design with the restrictions that WCAG has in place for contrast. At the same time the great designers love restrictions. So your reception will vary when you work with a designer. Remember, they don’t get a choice about adhering to the contrast levels set out.
Outlined in WCAG, we have two contrast levels we need to look out for text. We divide text into “normal” or “body” text and “large” text. Body text is all text smaller than 18 point and not bold. If it is bold, it can be as small as 14 point. Everything bigger than that is large text.
This is body text at 14 point.
This is large text, bold at 14 point.
This is large text at 18 point.
1 point does not equal 1 pixel.
1 point is actually 1.3 CSS pixels.
If you are like me, you are looking at those blocks thinking that they don’t look like 14 and 18 pixels. Heck, this body text looks the same size as the 18 point. Well that is because I use a larger than normal font for easier reading. Additionally, 1 point does not equal 1 pixel. 1 point is actually 1.3 CSS pixels. This means 14 point is really 19 rendered pixels. Eighteen point comes out to 24 pixels.
In Chrome and Firefox, the web inspector tools offer a “Computed” section. After highlighting the element in the code, switch to the computed tab and look for font-size. This will show the rendered pixel height regardless of what measurement you used to set the size. In addition to points, font sizes can be set with EM, REM, pixel, percentage, viewport height and width, just to name the most common.
Now that we know how to measure our fonts, let’s look at the contrast levels. We have two remember. One for body text and one for large text. For body text, the contrast ratio is 4.5:1. For large text the contrast ratio is 3:1. There are many contrast checkers on the market, so feel free to find one and use it.
My personal favorite is Level Access’s Accessible Color Picker for Chrome. Full disclosure, I currently work for Level Access. However, I’m not pitching the company. This color contrast tester not only provides the contrast and if it passes A, AA, or AAA, but is also helps guide you to fixing the colors.
On the side of the contrast values is a five by five grid. Your picked colors are in the middle. If you move up or down in the grid, options are shown for your background. Moving left or right in the grid changes the foreground. These options may have better contrast.
As a developer using this tool, I look for contrast pairs that don’t work. When I find an issue, I will suggest a color combo from these suggestions that does work. Then the designer just needs to approve, moving the process forward.
The WCAG is imperfect. It is a volunteer organization of hundreds trying their best to develop accessible rules the majority can agree on. On of the flaws is in the algorithm for finding contrast. There are combos that fall below WCAG standards that are no issue for people with color blindness. There are also combos that meet the criteria, but are unusable because they irritate eyes.
When the efforts to bring WCAG to the next level beyond 2.2, those involved began investigating other contrast algorithms. APCA is one of those. It looks like it will be how we evaluate contrast in the future. However, if you are building a website today it doesn’t matter. The current rules and laws all point back to WCAG 2.x and the algorithm used in that is not APCA. To comply with the current law, you must test colors under the current algorithm. This is true even if you use APCA to choose your colors. If you can use combos that pass both, you are future protecting yourself.
Have questions on this topic? Find me on LinkedIn and BlueSky to chat!
2 CommentsAlong with the <video> element, we also have an <audio> element that works very similarly. However, this particular success criteria isn’t talking about that, at least not exclusively.
Unexpected audio can be quite shocking for users. It doesn’t really matter if the music is a soft ambient melody. It also doesn’t matter if it’s a harsh explosive sound. It can cause issues for the user. For folks who have a sound sensitivity, not being able to control sounds causes problems. This can include persons with ADHD, Autism, hearing issues, vestibular disorders (my personal combo). They may have Hyperacusis (I do), but may not.
Hyperacusis is a disorder of the ear where every day sounds can become problematic. According to WebMD, it is considered “rare” at a rate of 1 in 50 thousand people have it. Tinnitus is a symptom, but you can have that without hyperacusis.
Hyperacusis comes in a few varieties. Loudness hyperacusis is a sensitivity that makes your head and ears feel full that can cause migraines. Pain hyperacusis (noxacusis) is a varietal that induces pain and discomfort to the user when the sound is heard. This often feels like stabbing or pins in the ears. Both occur when the volume of the sound is at a level considered “normal” for most people.
In addition to hyperacusis, your user may have a general intolerance to sounds humans often find annoying. These include car alarms, sirens, and barking, to name a few.
There’s also misophonia and phonophobia. For the most part these are anxiety based reactions that tend to be treated well using CBT style therapy.
Unexpected sounds can be quite problematic for folks. If the sound plays over 3 seconds it needs to have controls to pause or stop it. Avoid using unexpected sounds. Keep the volume reasonable. And it is always best to tie sounds to user action as it can be less shocking.
Have some thoughts? Want to discuss this further? Find me on Bluesky and LinkedIn to chat!
Comments closedColor is important. Color allows us to discern whether our fruit is ripe. We use color to control traffic. We spend time in school teaching kids their colors. We ask them to “Identify the red one.” But not all of us see color the same way, and some can’t see it at all. We cannot rely on it to be the sole means of communication.
Light hits an object, and the object applies a filter. This filter absorbs or reflects bands of the light spectrum. The filter is influenced by the materials used to make the object. This could be a pigment, but blue pigment is very rare in nature. When you see blue in animals, it is usually due to the structure not blocking blue light.
The eyes are made up of cones that absorb light. Humans have 3 cones, red, blue, and green. How much light hits each cone type also helps determine the color. Imagine a brand new tennis ball in full light glowing neon yellow. What happens if we cut the light in half? Is it still neon yellow? What if there is just enough light to see the ball? What color is it then?
Or was the tennis ball really neon green? Color blindness is the result of damage to the cones or them not fully developing. There are brain and optical nerve injuries that can cause it as well. To put it short:
My navy blue is not the same as your navy blue.
Trying not to put the horse before Decartes, but our brains are amazing machines. It takes in a extreme volume of stimuli and builds the world around us. While we teach each other a standard that allows us to participate in a society. But when you view blue (or any color), you don’t just get blue light, you get color influence. The colors surrounding your object will change how the colors appear.
These two circles are the same color. But they don’t look like it due to the backgrounds.
My wife has suffered several strokes and TBIs. We don’t see the same colors. It’s especially noticeable in the transition zones between colors. She asked me to grab something for her, but I couldn’t find it. She told me it was orange. Looking around I thought I found it, but it was red. She swears up and down it is orange. We’ve repeated this discussion with pinks, blues, and greens as well.
We each interpret colors differently. Therefore, it is vitally important that color is not the only way of communicating important information.
Like this article? Then please share. If you want to discuss it, find me on LinkedIn or BlueSky. If you liked my philosophical joke, leave a note and use the “polyphony.”
1 CommentWe have four main principles in accessibility: Perceivable, Operable, Understandable, and Robust (Accidentally wrote robot butts first!). The P.O.U.R. principles guide us in how we classify, identify, and fix issues.
When we provide a clear and concise label to an input, we move it from Perceivable to Understandable. Without a clear role, the user will not understand how to interact with it. When we use the same input for the same information in another place on the site, we must keep the naming consistent.
You will often see issues show up under SC 4.1.2 if they flag here. Fortunately, when inputs are named correctly, it resolves both issues. Inputs must have their proper role. Any additional data needed to operate them should be provided. What does this look like? Well it depends on the control type.
The <input> element offers up many alternatives. The default is the type of text. If you forget to include any type, the user will be able to type anything in. You should be familiar by now with radio buttons, checkboxes, and passwords. But we also have more specific versions like search, tel (telephone), URL, range, and color. When you use these native HTML versions, you are provided the the type of input for free. The browser and screen reader both recognize that they are special and behave a different way than type text.
It’s important to note a key point. While I’m discussing the details of the input element itself, the rules mentioned here apply to all active controls.
Labeling of an input can be handled multiple ways. The first way you should always attempt looks like this:
<label for="fieldID1">First Name:</label> <input type="text" id="fieldID1" />
Simple, native HTML. Beyond connecting the label to the field for the screen reader, we assist other users. If a user has issues with their fine motor skills, they don’t need to click on the field alone. The label will place focus in the field. This is super handy for radios and checkboxes due to their size. Further, voice control users will be able to see the label. They can then say “Click First name” and focus will go there.
But this isn’t the only way to do it. If we venture off the above method, we need to make sure we bring the right ARIA along with us.
Along with the ARIA, we want to use autocomplete. The concept of autocomplete has been around for a while. We have browsers that help us save important information. We have password managers the do the same thing more securely. These are wonderful things! They help all users get to the end goal faster. So why doesn’t everyone use autocomplete?
For the majority, it is a security issue. They prevent copy and paste on the input field forcing you to recall the information. But I’m not the first to tell you how hard it is to remember them. Which is why people use the same passwords everywhere and are easy to hack.
One of the ideas about accessibility that people often forget is that accessibility is security. Let’s imagine you are blind and need cash now (J.G. you out there?). You find an ATM, but there is no Braille, no headphone jack, and it is touchscreen only. If the world has been graceful, they are with a confident they can trust with their private information. But if they are solo, who do they ask? Anyone becomes a security threat then.
This same user arrives at work, sits at her desk and boots up the computer. Headphones on, and login, but she isn’t aware that someone has shoulder surfed her password.
Heading back home on the bus, she takes out her phone to buy more dog food online. She tries to input her credit card info. Due to poor labeling, she inputs the card number in the wrong box. She then asks a stranger for help…
When you allow users to paste and use autocomplete on their end, you help those with disabilities. And the security gaps get smaller as the user is more independent.
Have questions or challenges with this topic or my position? Let’s talk them out on LinkedIn or BlueSky!
1 CommentIt’s not the 80’s. We don’t live in a world where “the fold” exists any longer. We’re not locked to a 4:3 viewing experience. Our movies are most commonly 16:9 ratio. Our TVs and phones have followed this pattern. In short, we need to pay attention to how our sites look on various screen dimensions. We won’t know what the viewer is using.
We need to be flexible in our designs. Most of the time when we design a website or application we do so with a primary orientation. For all devices other than tablets and phones, this is in landscape mode. But for phones and tablets we primarily design for the portrait mode.
There are times when a user may have their device stuck in a position that they are unable to change. When this situation occurs, if our design is locked into the Portrait orientation, we make it very difficult for some users. It would be much more usable in landscape for this moment or by the user. Therefore we should make it work in both ways.
The primary example of this is someone using a mounted iPad on their wheelchair. They could have it in portrait mode, yet not have the strength to turn the device.
But of course! There are certain tasks that the orientation can’t change. Usually you hear about digitally depositing checks. The thinking here is using the phone in landscape mode is required to get the right photo. This is a crap example. I can just as easily take that photo in landscape as long as I can square it in the view. A better example might be needing you to turn your phone landscape to read and interact with a chart. Another would be watching a video. If shot for mobile device apps like TikTok, it doesn’t work as well in portrait.
Mobile platforms have the ability to do dynamic layouts now. These layouts allow the user to change their display orientation without affecting their use of the product. However, using a dynamic layout requires you start coding with that in mind. Trying to fix the layout after the fact is much more difficult.
Have some thoughts you’d like to discuss? Maybe other exception examples? Reach out to me on BlueSky or LinkedIn.
1 Comment
