Skip to main content
WCAGrules
Quick navigation

The A to Z glossary

Accessibility, translated. Every term, A to Z.

131 terms, from screen reader and alt text to focus trap and Level AA, explained the way we would say them across a desk rather than the way the spec writes them. Most of what makes an accessibility report hard to read is vocabulary, not logic. So each entry gives you the plain version first, then sends you on to the rule, the fix, the law or the free tool where the word actually costs you something.

Terms starting with A

Top ↑

a11y

a11y is shorthand for the word accessibility, and it means that and nothing narrower. The number is a letter count. Take accessibility, keep the a and the y, and replace the 11 letters in between with the number. That trick has a name, numeronym, and you will meet its siblings all over software. i18n for internationalization, l10n for localization, k8s for Kubernetes. Out loud you will most often hear "ally", which is worth knowing before you sit in a meeting and try to pronounce it off the page.

a11y, in full →

On this siteAccessibility · What web accessibility means

Abmahnung

Also called: German cease-and-desist warning letter

An Abmahnung is a cease-and-desist letter from a competitor rather than from a regulator, and in Germany it is the accessibility risk actually landing on doormats. German unfair competition law lets a rival demand that you stop breaking a rule that gives you an edge over them, and the demand turns up with the sender's legal costs attached and a pre-written promise for you to sign. That promise is the expensive half rather than the invoice, because it binds with no end date and carries a penalty that falls due the next time the same fault appears on your site. The law is stricter with the sender than most people expect. The letter has to state five specific things to be worth anything, and where it does not, the business that received it can claim its own defence costs back, up to the amount that was demanded.

Abmahnung, in full →

On this siteGermany's BFSG · The European Accessibility Act · European Accessibility Act audit

Accessibility

Accessibility means building websites and tools so that people with disabilities can use them. W3C puts it more precisely, as being able to perceive, understand, navigate and interact with the web, and to contribute to it. That last verb is the half most sites forget. A page can be perfectly readable and still carry a comment box, a review form or an upload that nobody using a screen reader can complete. The range it has to cover is wider than the obvious cases too, taking in auditory, cognitive, neurological, physical, speech and visual disabilities. And it is measured rather than felt. The measure is WCAG, the level laws and contracts reach for is almost always AA, and in a great many countries some version of this is required by law.

Accessibility, in full →

On this siteEvery WCAG rule, explained · The laws that require it · Accessibility, usability and inclusive design

Accessibility API

An accessibility API is the channel an operating system provides so assistive technology can find out what is on screen. Your browser reads the page, builds an accessibility tree out of the markup, the CSS and any ARIA on top, and hands that tree to the platform. A screen reader asks the platform rather than reading your HTML, and ARIA calls that the preferred route while permitting software to read the markup out of the DOM instead. Every desktop and mobile platform has one, and the names differ. Windows has two in active use. UI Automation, and the older Microsoft Active Accessibility with its IAccessible2 extension, which is what JAWS and NVDA actually consume from Firefox and Chrome. On macOS it is the AX API, which a Mac developer meets under the name NSAccessibility. Android has its own, and Linux splits the job in two, between an interface applications implement and a desktop service that gathers it for the tools reading it.

Accessibility API, in full →

On this siteThe accessibility tree · Rule 4.1.2 Name, Role, Value

Accessibility audit

An accessibility audit is a structured evaluation of a website against WCAG, ending in a documented pass or fail for every applicable rule with the evidence attached. W3C publishes a method for it, and testing is only one of five steps. You define the scope, explore the product, select a sample, evaluate that sample, then report what you found. The two steps before testing are where cheap audits actually go wrong, because a sample chosen badly makes everything after it unreliable. A real audit also uses more than one kind of pass. A machine finds what machines can decide, an expert reviews the judgment calls, and somebody who uses assistive technology daily works through the journeys that matter.

Accessibility audit, in full →

On this siteHow we run an audit · Free 10-page scan

Accessibility statement

An accessibility statement is a public page saying how accessible your site actually is. Three things make it worth reading. It names the standard you are aiming at, usually WCAG at Level AA. It says what is not accessible yet and what somebody can do instead. And it gives a person blocked by a barrier a real route to a human, with an answer they can expect inside a stated time. The wording W3C models for the middle part is partially conformant, meaning some parts of the content do not fully conform. That is the honest position for very nearly every site, and it is a far stronger place to stand than a claim you cannot back.

Accessibility statement, in full →

On this siteGenerate yours free · Which laws require one

Accessibility supported

A way of using a technology is accessibility supported when two things are true together. First, that way of using it has been tested for interoperability with the assistive technology people actually have, in the human languages the content is written in. Second, there is an accessibility-supported user agent people can get hold of, which means either widely distributed software, or a widely distributed plug-in, or a closed environment like a corporate network where the required software is supported, or a user agent available to download or buy that costs a disabled person no more than anybody else and is no harder for them to find and obtain. Something can be valid, documented and specification-correct and still fail both halves.

Accessibility supported, in full →

On this siteThe conformance requirements · ARIA explained · Screen reader testing

Accessibility tree

The accessibility tree is a second structure the browser builds alongside the DOM, holding one object for every part of your page that has to be exposed to assistive technology. Four things on each object do most of the work. Its role, meaning what it is. Its accessible name, meaning what it is called. Its description, meaning anything extra attached. And its state. Pressed, expanded, checked, disabled. There is more on there besides, including how objects relate to one another and what can be done to them, so treat those four as the ones you will be fixing rather than the whole record. Assistive technology works from that tree, by way of the operating system's own accessibility interface, rather than from your CSS or your markup directly. So if the tree is wrong, the page is broken for those readers however good it looks on screen.

Accessibility tree, in full →

On this siteRule 4.1.2 Name, Role, Value · ARIA explained

Accessible description

Also called: aria-describedby, aria-description

An accessible description is the extra line a screen reader reads out after the name, and the word the specification uses for what it does is complements. The name says what a control is for. The description says what else somebody needs to know before they use it, and nothing in a description can touch the name. That one rule is why aria-describedby is the safe way to attach context to a button and aria-label is not. Four sources feed the description and the first one that applies wins, in the order aria-describedby, aria-description, a small set of HTML elements, then the title attribute. First match wins even where the match is empty, which is the trap in it. Point aria-describedby at an id that is not on the page and you get no description at all, and the title attribute that would have worked is never reached.

Accessible description, in full →

On this siteHow a description is computed · How a name is computed · aria-describedby

Accessible name

The accessible name is what assistive technology announces an element as, the words a screen reader speaks when it lands on a button or a link. It is computed rather than declared, from the text inside the element, a label, alt text, or ARIA attributes, and the order they are consulted in is what surprises people. It runs aria-labelledby first, then aria-label, then whatever the element's own technology supplies such as a label or alt text, then the text inside, and title last of all. So an aria-label on a button beats the words printed inside it, while a title loses to almost everything and is a fallback rather than an override. Some elements cannot be named at all, because their role forbids it, which is why the attribute sometimes does nothing whatever. And an element with no name is announced as its role and nothing else. "Button." "Link." "Edit text." The reader knows something is there and has no idea what it does.

Accessible name, in full →

On this siteRule 4.1.2 Name, Role, Value · Rule 2.5.3 Label in Name

ACR

Also called: Accessibility Conformance Report

An ACR is a completed VPAT. The template arrives blank, you test your product against each requirement and record the result, and the filled-in document is called an Accessibility Conformance Report. Buyers ask for a VPAT and mean this one. What they receive is a self-declaration rather than a certificate, because nobody accredits an ACR and it is worth exactly the testing that produced it. The edition matters as much as the answers do. There are four of them and they carry different versions of WCAG. A report on the Section 508 edition assesses the web criteria against WCAG 2.0, alongside the software and support-documentation chapters that are not WCAG at all. The EU edition reports 2.1, and only the WCAG and INT editions reach 2.2.

ACR, in full →

On this siteVPAT and ACR, explained · VPAT preparation · What a conformance report cannot say · A customer asked for a VPAT · Whether a report covers your configuration

ADA

Also called: Americans with Disabilities Act

The ADA is the main US disability rights law, signed in 1990 and broadened by amendments in 2008. It never mentions websites, which has slowed nothing down, because the Department of Justice has long treated what a business offers online as covered. The courts are less settled than that sounds. Where a site is the online face of a business with premises, claims generally proceed. Where the business exists only online, the answer changes with the circuit, and no federal appeals court has yet held that a website on its own is a place of public accommodation. Which part of the Act applies to you decides almost everything after that. Title III covers businesses open to the public, and no regulation there sets a technical standard, so you choose how you make your online services usable and WCAG is what everybody reaches for. Title II covers state and local government, and there WCAG 2.1 Level AA is now written into a federal rule with dates attached.

ADA, in full →

On this siteThe ADA, explained · WCAG vs. the ADA

alertdialog

Also called: Alert dialog

An alertdialog is a dialog that interrupts, and it is the role for a message somebody has to answer before anything else happens. ARIA builds it out of two parents at once. From alert it takes the urgency, so the browser is meant to fire a system alert event the moment the dialog is drawn, wherever the platform has such an event. From dialog it takes the ability to be answered, which a plain alert never has. Three obligations come with the role. It needs an accessible name, and that name has to come from aria-label or aria-labelledby rather than from the words inside it. Its message belongs in aria-describedby, pointed at the element holding those words, because a screen reader left to work the message out for itself is guessing. And it should be modal, meaning nothing outside it answers the keyboard or the mouse while it is open. Reach for it when unsaved work is about to go, and use a plain dialog for everything else.

alertdialog, in full →

On this siteThe alertdialog role reference · Accessible modals and dialogs · aria-describedby · Why a chat error in an alertdialog goes unheard

Alt text

Also called: Alternative text, alt attribute

Alt text is the written description attached to an image, and it is what a screen reader reads out in place of the picture. The test is not whether you described it. It is whether the reader gets what the image was there to give them, which is why the same photograph needs different alt text on different pages. A chart's alt text is the finding. A product shot's is the product. An image sitting alone inside a link is a special case, because there the words have to name where the link goes rather than what the picture shows. And decoration gets an empty alt attribute, which tells assistive technology to skip it.

Alt text, in full →

On this siteRule 1.1.1 Non-text Content · Alt text checker · What to write, in four questions

ARIA

Also called: WAI-ARIA, Accessible Rich Internet Applications

ARIA is a set of attributes that tell assistive technology what a custom control is and what state it is in. ARIA 1.2 defines 48 of them, every one named aria-something, plus the role attribute, whose value comes from a vocabulary of 94 roles. Twelve of those roles are abstract and exist only to hold the vocabulary together, so authors must not put them in content and the number you can actually write is 82. It exists for the things plain HTML has no element for, like tab strips, tree views and live announcements. What ARIA cannot do is change behavior. Put a button role on a div and a screen reader announces a button, while pressing Enter still does nothing, because you promised a button and shipped a div. That is W3C's own framing. A role is a promise, and the first thing its authoring guidance says is "No ARIA is better than Bad ARIA".

ARIA, in full →

On this siteRule 4.1.2 Name, Role, Value · ARIA techniques, explained

ARIA Authoring Practices Guide

Also called: APG

The ARIA Authoring Practices Guide is W3C's pattern library for interface widgets, and the index runs to thirty. Tabs, tree views, comboboxes, dialogs, carousels. Some are things HTML has no element for. Others are things it does, because people rebuild those by hand anyway. Each one gives the markup, the roles, the states and, above all, the keyboard behaviour people expect, because browsers supply none of that for a control you built yourself. It is where W3C now sends readers for this, since Using ARIA, the document that used to carry the rules of ARIA use, has been discontinued and points here instead. Two limits are worth carrying in with you. The sample code is written to demonstrate the specification rather than to survive contact with real assistive technology, so W3C tells you to test anything you copy. And thirty is fewer than it sounds, because the pattern you came looking for is often not one of them.

ARIA Authoring Practices Guide, in full →

On this siteARIA explained · ARIA techniques

Assistive technology

Also called: AT

Assistive technology is hardware or software that gives a person with a disability more than a mainstream browser offers on its own. It either acts as the browser itself or works alongside one, and WCAG names four things it does. It presents content another way. Speech, or magnified text. It accepts another kind of input, like voice. It adds ways to navigate, which is what a screen reader's headings list really is. And it transforms content, which covers the reading and simplification tools people with cognitive disabilities rely on. Your site never talks to it directly. The browser builds a description of your page, the operating system passes that on, and the tool reads what arrives.

Assistive technology, in full →

On this siteHow we test with real AT users · What people actually use

Audio description

Audio description is narration added to a video's soundtrack that describes the visual detail the soundtrack alone does not carry. Actions, characters, scene changes, text on screen. In its standard form it goes into the pauses that already exist in the dialogue, which is why a video with wall-to-wall talking is harder to describe than a quiet one. It turns up at all three WCAG levels rather than only one, and every one of the three is about prerecorded video with a soundtrack. At Level A you can choose between a full text alternative and description. At Level AA the choice disappears and the description itself is required. At Level AAA you owe an extended version, where the video pauses to make room when the natural gaps are too short.

Audio description, in full →

On this siteRule 1.2.5 Audio Description · Video, captions and transcripts

Autocomplete

Autocomplete is an HTML attribute naming what a form field collects. Name, email, tel, street-address, postal-code and 48 others. It lets browsers and password managers fill fields for people. It also lets assistive software show a familiar icon beside a field instead of a word, which works whatever language the label is written in. Rule 1.3.5 is the WCAG requirement behind it at Level AA, and it asks for something slightly different from what most people think. The purpose of the field has to be programmatically determinable. The attribute is the way you do that in HTML rather than the rule itself, so what fails is a field with no declared purpose, not a field where autofill has been switched off. The rule also only reaches fields collecting information about the user, so a delivery address for somebody else is outside it.

Autocomplete, in full →

On this siteRule 1.3.5 Identify Input Purpose · The autocomplete technique · All 53 input purposes, listed

Automated testing

Also called: Accessibility scanning

Automated testing is software that checks a page's code for the accessibility failures a machine can prove. An image with no alt attribute. A form field with no label. Text measuring below the contrast threshold. Run it, because it is fast, it is cheap, and what it measures it measures the same way every time. What it cannot do is decide conformance, and that is not a gap in the tools. W3C says outright that no tool alone can determine whether a site meets the standard, and that knowledgeable human evaluation is required. So passing the scanner and being accessible are two separate claims, and only one of them is a thing a machine can settle.

Automated testing, in full →

On this siteRun our free scanner · What machines can and cannot check

Terms starting with B

Top ↑

Braille display

Also called: Refreshable braille display

A braille display is a hardware device with cells whose pins rise and fall to form braille characters, refreshing as the reader moves through a page. Nearly all of them are a single row. It is an input device as well as an output one. Most carry a braille keyboard, and each cell has a key above it that moves the cursor to that character, which is the braille equivalent of clicking. It takes its text from a screen reader, so anything that breaks the screen reader breaks braille in the same moment. On a single-line display, text is all that reaches the reader. No color, no layout, no image, and nothing that lives only in your CSS. Multi-line devices that raise tactile graphics as well now exist, and they are the exception rather than what your reader is likely to own.

Braille display, in full →

On this siteHow screen-reader testing works · Assistive technology, in full

Browse mode

Also called: Forms mode, focus mode, scan mode, the virtual cursor

Browse mode is the state a Windows screen reader sits in while somebody reads your page, and forms mode is the state it switches into when they type. In browse mode the screen reader takes a flat copy of the page and drives its own reading cursor through it with the arrow keys, which frees every letter on the keyboard to work as a shortcut. H jumps to the next heading, B to the next button, F to the next form control. In forms mode almost every key goes straight to the control instead, because an edit box has to be able to receive an H. Each product names the pair differently. NVDA says browse mode and focus mode, JAWS calls its reading cursor the Virtual Cursor and its typing state Forms Mode, and Narrator calls the reading state scan mode. All three switch on their own most of the time, and that switch is why a keyboard test and a screen reader test turn up different bugs on the same page.

Browse mode, in full →

On this siteHow to test with a screen reader · Rule 2.1.4 Character Key Shortcuts · The application role, which switches browse mode off · What a reader misses inside a grid

Bypass blocks

Bypass blocks is rule 2.4.1, and it asks for one thing at Level A. Somewhere there has to be a mechanism that gets a person past content repeated across pages, which in practice means the navigation. A skip link is the answer everybody reaches for and it is one of eight, because W3C offers two routes with several techniques under each. You can link past the block, or you can group the block so it can be skipped, and landmark regions or a real heading structure do the second on their own. It is narrower than it sounds, too. Individual words, phrases and single links are not blocks. And browsers already provide a keyboard route back to the top of a page, so where the navigation sits in the footer W3C says a skip link there may be unnecessary.

Bypass blocks, in full →

On this siteSkip link · How to add a skip link

Terms starting with C

Top ↑

CAPTCHA

Also called: Completely Automated Public Turing test to tell Computers and Humans Apart

A CAPTCHA is a test a site puts in front of you to prove you are a person and not a program. Distorted letters to retype, or photos to pick the traffic lights out of. WCAG does not ban them, and what it does ask for is more than most sites provide. Two things are required together, not one or the other. Text identifying and describing what the test is for, and alternative forms of it using different senses, so somebody who cannot do the visual version has a real route through. WCAG 2.2 then added a tighter rule for logins. A cognitive test is not allowed there at all unless the step also offers one of four things. Another way in, a mechanism that helps you complete the test, object recognition, or something you provided yourself.

CAPTCHA, in full →

On this siteRule 1.1.1 Non-text Content · Cognitive accessibility · Where a CAPTCHA sits in a sign-in audit

Captions

Also called: Closed captions, subtitles for the deaf and hard of hearing

Captions are a synchronized text version of everything you would hear in a video, covering speech and the non-speech sound that matters as well. Sound effects, music, laughter, and who is speaking. Speaker identification is the one to check for, because a conversation between three people is unfollowable without it and a caption file can look complete while carrying none. Closed captions are the kind the viewer can turn off, which usually means a separate file the player reads, and that separation is what lets a player offer resizing, restyling or another language where it supports them. Open captions are burned into the picture and can do none of that. The standard accepts either. Prerecorded video needs them at Level A, and live video needs them at Level AA, where there is no exception clause at all.

Captions, in full →

On this siteRule 1.2.2 Captions (Prerecorded) · Rule 1.2.4 Captions (Live)

CART

Also called: Communication Access Realtime Translation, realtime captioning

CART is live captioning written by a person rather than a machine, and the letters stand for Communication Access Realtime Translation. A trained captioner writes on a steno machine into realtime software, and the words land on a screen, a laptop, a tablet or a phone a second or two behind the speaker. What you are buying is accuracy. The service is word for word by definition and it carries environmental sound as well as speech, because a caption that drops the fire alarm has not described the room. The credential to ask a supplier for is Certified Realtime Captioner, which is passed by writing at 180 words a minute at 96% accuracy. That pair of numbers is the thing to hold a quote against, and it is why automatic speech recognition is a different purchase rather than a cheaper version of the same one. Most CART now arrives remotely, so the captioner does not have to be in the room.

CART, in full →

On this siteRule 1.2.4 Captions (Live) · Captions and transcripts · Video and media review

Change of context

Also called: Context change

A change of context is one of four things, and only four. The user agent changes, the viewport changes, focus moves, or the content changes in a way that changes the meaning of the page. Everything else your interface does when somebody clicks is a change of content, which is a different thing under different rules, and most of a modern interface lives there. No entry in WCAG's glossary carries more of the standard than this one. Three success criteria name it and either forbid it outright or make you warn about it first, and a fourth is scoped by it from the other side, because a status message is defined as a change in content that is not a change of context. The member people trip over is focus, since moving focus feels like a courtesy rather than an interruption and the definition counts it on its own terms. That is why a confirmation which announces itself and also takes focus gets read out twice.

Change of context, in full →

On this siteChange of context, in full · Rule 3.2.1 On Focus · Rule 3.2.2 On Input · Rule 4.1.3 Status Messages

Cognitive disability

Cognitive disability is a broad group of conditions affecting how people take information in, hold on to it, and act on it. It covers reduced ability in areas like reading, writing, communication or number work, difficulty understanding new or complex information, and reduced memory and attention. Intellectual disabilities, dyslexia, ADHD, autism, brain injury after a stroke or a concussion, early-stage dementia, and mental health conditions such as anxiety all sit inside it, and W3C's own scope for this area names neurodiversity alongside them. One thing is worth saying plainly, because it is the most common misconception about the whole category. Overall intelligence is often not affected at all. Everybody has bad days with a confusing interface, and what makes this a disability is that the difficulty is persistent and significant enough to stop somebody finishing the task on their own.

Cognitive disability, in full →

On this siteThe Understandable rules · Cognitive accessibility

Color blindness

Also called: Color vision deficiency

Color blindness is a reduced ability to tell certain colors apart, and the figure everybody quotes, roughly 1 in 12 men and 1 in 200 women, is for red-green deficiency among people of Northern European descent. Prevalence is lower in other populations, so read it as an order of magnitude rather than a headcount for your audience. The common forms make red and green hard to separate, which is why red error states, green success states and red-and-green chart series are the classic failures. It reaches further than the name suggests, though. Red-green deficiency also confuses blue with purple, because purple has red in it. A second and rarer form runs on blue and yellow instead, so switching your chart to those is not the safe answer people assume. Seeing no color at all is rare. The rule that follows from all of it fits on one line. Never make color the only way to tell something.

Color blindness, in full →

On this siteRule 1.4.1 Use of Color · Colour and contrast

Conformance

Conformance means satisfying all the requirements of a standard, and WCAG has five of them rather than one. The first is the level. Every applicable success criterion at A, AA or AAA is met in full, or a conforming alternate version is provided instead. The other four are not success criteria at all, and none of them comes in separate A, AA and AAA versions. They are applied at whatever level you are claiming. The whole page counts and no part of it can be set aside. Every page in a process conforms or none of them does. Only accessibility-supported ways of using a technology can be relied on. And anything you are not relying on still must not block access to the rest of the page. The other thing to know is the unit. WCAG defines conformance for web pages only, so it is always a page that conforms and never a site in the abstract. What a claim can cover is one page, a series of pages or many related pages, and that is the sentence people mean when they say the site is AA.

Conformance, in full →

On this siteLevel AA: the legal standard · WCAG versions and levels

Conformance claim

A conformance claim is an optional formal statement that content meets WCAG at a stated level. Make one and it has to carry five things. The date. The guidelines title, version and address, written out rather than implied. The level satisfied. A concise description of the pages covered, which has to say whether subdomains are included in the claim. And a list of the web content technologies relied upon. A conformance logo is itself a claim, so a badge in a footer with none of that behind it is an incomplete claim rather than a decoration.

Conformance claim, in full →

On this siteThe conformance requirements · Writing an accessibility statement · VPAT and ACR

Conforming alternate version

A conforming alternate version is a second version of a page that does meet WCAG, standing in for one that does not, and it is a route through the standard rather than a workaround. Conformance at any level is satisfied either by the page itself meeting the criteria or by an alternate version that does. Four requirements come with it. It has to conform at the level being claimed. It has to carry the same information and the same functionality, in the same human language. It has to be as up to date as the content it stands in for. And one of three things has to be true about how the two versions connect. Either the conforming version can be reached from the non-conforming page by a mechanism that is accessibility supported. Or the non-conforming version can only be reached from the conforming version. Or it can only be reached from a conforming page that also offers a way to the conforming one.

Conforming alternate version, in full →

On this siteUniversal design · Conformance

Contrast ratio

A contrast ratio is a number from 1:1 to 21:1 measuring how far apart two colors are in brightness, computed from a formula rather than judged by eye. Body text needs 4.5:1 at Level AA. Large text needs 3:1, where large means 18 point, or 14 point bold, which works out at roughly 24px and 18.5px. A separate rule asks 3:1 of non-text things, and its scope is narrower than the shorthand suggests. What it covers is the visual information somebody needs for identifying a control or its state, and the parts of a graphic somebody needs for understanding it. So the boundary of a text field and the focus ring around it are in, and an icon sitting beside a label that already says the same thing is not. The formula is why light gray on white keeps failing audits. It can look perfectly readable to a designer with good sight and still fall short, and the threshold is never rounded, so 4.49:1 fails 4.5:1.

Contrast ratio, in full →

On this siteRule 1.4.3 Contrast (Minimum) · Check a color pair · Build a passing palette

Terms starting with D

Top ↑

Dark mode

Also called: prefers-color-scheme

Dark mode is a second colour scheme with a dark background. You offer it by honouring what the visitor has set in their operating system or browser, or by a toggle on the page, or by both. It is a separate palette rather than an inversion of the first one. Whichever scheme somebody lands in is the presentation they get, so the pairs in it have to clear the same thresholds the light ones do. WCAG scopes contrast to the colour pairs an author would expect to appear together, and a palette you wrote is a presentation you caused. There is a route for a scheme that deliberately falls short, and it is the alternate-version one, which means a conforming scheme has to stay reachable and the switch between them has to work for everybody. Nothing in WCAG requires a dark mode at all. What WCAG does require is that the one you ship works, which is a bigger job than it looks, because a second palette doubles the surface every contrast rule applies to.

Dark mode, in full →

On this siteColour and contrast · Check a colour pair · Why an untested theme stays untested

Data visualisation

Also called: Chart, graph, infographic

A data visualisation is any graphic that turns numbers into a picture. To WCAG it is non-text content like any other, so what it needs is a text alternative that serves the equivalent purpose, and that phrase is the whole argument. The purpose of a chart is the finding, so an alternative that names the object has not served it. For something this complicated the working pattern is two-part. A short description saying what the chart is and where the longer one lives. Then a longer description carrying the information the chart conveys, which usually means the numbers themselves on the page as a real table or a paragraph. The failure has a name in W3C's own list. A text alternative you could not put in place of the image without losing something is not an alternative.

Data visualisation, in full →

On this siteAccessible charts · Dashboard and data viz audit · Auditing an authored Power BI report

Decorative image

Also called: presentational image

A decorative image is one that serves only an aesthetic purpose, carries no information, and does no work. All three conditions have to hold, because that is the standard's own test rather than a judgment about how pretty something looks. Dividers, background textures, and an icon sitting beside a word that already says the same thing all qualify. The third condition is the one that catches people, since an icon inside a link is not decoration when it is the only thing naming that link. Where an image really is decoration, the rule does not ask you to describe it. It asks you to mark it up so assistive technology can skip it, and an empty alt attribute is the usual way of saying so.

Decorative image, in full →

On this siteAlt text · Does this image need alt text? · 1.1.1 Non-text Content

Disability

Disability results from the interaction between a health condition and the world around it. That is the World Health Organization's framing, and it is the one that matters for web work, because it puts the barrier rather than the person at the center of the question. Somebody who is blind is not stopped by blindness on a well-built site. They are stopped by an unnamed button. WHO estimated in March 2023 that 1.3 billion people experience significant disability, about 16% of the world, or one in six. The categories web accessibility has to reach are auditory, cognitive, neurological, physical, speech and visual. Speech is the one that falls off the list soonest, which is how voice-only support lines keep getting built.

Disability, in full →

On this siteWhat WCAG asks you to do about it · Why accessibility matters

Disabled Access Credit

Also called: ADA tax credit, IRS Form 8826

The Disabled Access Credit is a US federal tax credit that takes half of what an eligible small business spends removing barriers for people with disabilities off its tax bill. It covers 50% of eligible spending above $250 and up to $10,250 in a year, so the largest credit anybody can claim is $5,000. It is nonrefundable and it runs through the general business credit rules, which means it reduces tax you owe rather than arriving as a cheque. You are eligible if, for the previous tax year, your gross receipts were $1,000,000 or less, or you had 30 or fewer full-time employees. Either test on its own is enough and both bounds are inclusive, so exactly 30 employees still qualifies. You claim it on IRS Form 8826, and it resets every tax year. Whether website accessibility work counts as eligible spending is an open question, so treat accessibility audit costs as costs that may qualify. Ask your accountant.

Disabled Access Credit, in full →

On this siteWhat the audits cost · The ADA, explained

DOM

Also called: Document Object Model

The DOM is the browser's live, in-memory version of your page, a tree of objects it builds by parsing your HTML and then keeps updating as your scripts run. It is what you see in developer tools, and it is usually not what your server sent. That gap is the whole point of the term. Content a script adds after load exists in the DOM and nowhere in your source. That is how an overlay widget changes what a scanner sees while your actual markup stays exactly as broken as it was. Assistive technology mostly does not read the DOM directly. The browser derives the accessibility tree from it, the operating system exposes that, and the screen reader reads whatever arrives at the end of the chain. ARIA calls that the preferred route rather than the only one, and allows software to read the markup straight out of the DOM instead, which is how a tool that reformats a page for easier reading works.

DOM, in full →

On this siteAccessibility tree · ARIA explained

Dyslexia

Dyslexia is a set of processing difficulties that affect the acquisition of reading and spelling. Reading fluency and spelling are the markers, rather than anything to do with the eyes. That distinction matters, because the popular account of dyslexia is a visual one and the clinical consensus is not. The difficulty underneath it is most often phonological, which is the mapping between sounds and letters. It does not affect intelligence. It sits on a continuum rather than being something you either have or do not. And it rarely travels alone. Attention differences, dyscalculia, language disorder and coordination disorder co-occur with it often enough that the combination is the norm.

Dyslexia, in full →

On this siteRule 1.4.12 Text Spacing · Fonts and typography

Terms starting with E

Top ↑

Easy read

Easy read is a document format made for people with learning disabilities. Simple text, free of jargon and cut back to the information that matters most, with clear images beside it to support the meaning. UK government guidance is blunt about what it is not, saying in one sentence that easy read is different from plain English. So it is a separate discipline rather than plain English turned up a notch. Europe treats it as a defined format too, with published standards called Information for all and an official logo a conforming document can carry. The part organisations get wrong is the scope. People with learning disabilities need access to all of your information, not only the parts that are about disability.

Easy read, in full →

On this sitePlain language · Cognitive disability

EN 301 549

EN 301 549 is the European standard for accessible information and communication technology, written first for public procurement and now the document European accessibility law reaches for when it needs a technical answer. For a website the operative part is clause 9. It adopts WCAG 2.1 at Levels A and AA outright rather than restating them, and adds WCAG's five conformance requirements in a clause of its own, so most of what the standard asks of a site is WCAG AA under a different number. The rest of it covers ground WCAG never touches, including hardware, software, non-web documents, real-time text inside a voice call, and one clause about your own documentation and support channels. Version is the thing to watch here. The published version in force is V3.2.1, which is where the WCAG 2.1 figure comes from. A successor moving to WCAG 2.2 has been drafted and is not yet cited in the Official Journal. So it is not the legal yardstick anywhere yet.

EN 301 549, in full →

On this siteThe European Accessibility Act · VPAT and ACR, explained

EPUB accessibility

EPUB accessibility is what an e-book has to do to be readable by everyone who buys it, and the standard that says so is EPUB Accessibility 1.1, a W3C Recommendation dated 17 October 2024. It does two jobs rather than one. It sets conformance requirements for the content. And it sets metadata requirements, so a reader can find out whether a book works for them before opening it, which is something WCAG has no equivalent for. The floor is lower than most people assume. A publication has to meet WCAG 2.0 at Level A, and the latest version of WCAG at Level AA is only strongly recommended. So a title can conform to the standard and still fall short of what a contract or a law is asking for.

EPUB accessibility, in full →

On this siteThe European Accessibility Act · PDF/UA

Error identification

Also called: Error messages

Error identification is the Level A rule that when your form rejects something, the page names the field that failed and says in text what went wrong. Both halves matter. The second is where most forms fall down. A red outline around the box carries the message in a colour some people cannot see, and it says nothing at all to somebody listening to the page. What the rule does not do is tell you where to put the message. A summary at the top of the form conforms as well as a line under the field, so the failure is almost never the placement. It is "Something went wrong" being the entire message. A second rule at Level AA then asks you to suggest the fix where the error was detected automatically and a fix is known, unless saying so would put the security or the purpose of the content at risk. A well-written message usually does both jobs in one sentence.

Error identification, in full →

On this siteRule 3.3.1 Error Identification · Rule 3.3.3 Error Suggestion

Essential

Essential is the word sixteen success criteria use to carve out the cases where meeting them would change what is actually on offer. It is not a general escape hatch, and it is not available on every rule. A criterion either carries the word in its own text or it does not, so the first question is always whether the rule you want to set aside has an essential clause at all. Where it does, the bar is written into the definition and has two halves. Removing the thing you want excused would have to fundamentally change the information or the functionality, and that same information and functionality has to be unreachable any other way that would conform. Presentation is not automatically outside it either, since the images-of-text rule names a particular presentation as capable of being essential, which is how a logotype survives.

Essential, in full →

On this siteRule 1.3.4 Orientation, which carries the exception · Rule 1.4.5 Images of Text, which carries it too · The conformance requirements

European Accessibility Act

Also called: EAA

The European Accessibility Act is the EU law that has required certain consumer products and services to be accessible since 28 June 2025, and the word to hold onto is certain. It is not a general web accessibility law for Europe. It covers a closed list. The entries a website owner cares about are e-commerce, consumer banking, e-books, telecoms, parts of passenger transport and access to audiovisual media. A site is in scope where it delivers one of those, not because it exists. Where you are covered, being outside the EU does not help, because what counts is selling to consumers in the EU rather than where you are established. Where you may not be covered at all is if you are a microenterprise providing a service. That means fewer than 10 people, and turnover or a balance sheet total no higher than two million euros. The Act names no WCAG version. It sets functional requirements and grants a presumption of conformity to harmonised standards, and no standard has been cited for it yet. So a website goes to EN 301 549 for its technical answer, and from there to WCAG 2.1 Level AA, which is the best reading available rather than a presumption anybody can claim.

European Accessibility Act, in full →

On this siteThe EAA, explained · The report buyers ask for

Terms starting with F

Top ↑

Focus

Also called: Keyboard focus

Focus is the one place on a page that is currently taking keyboard input. Type, and the characters go there. Press Enter or Space, and whatever is focused is what responds. Only one thing has it at a time, which is why moving focus is how anybody driving the page without a mouse gets from one control to the next. Two things about it are worth having straight early. The rule underneath is about functionality being operable through a keyboard, not about Tab reaching every element. A well-built menu that you enter with Tab and then move through with arrow keys is the correct pattern, not a gap. And a screen reader has its own reading cursor that moves independently of focus. Where the screen reader is and where focus is are two different questions. They only sometimes have the same answer.

Focus, in full →

On this siteRule 2.1.1 Keyboard · Keyboard accessibility

Focus indicator

Also called: Focus ring, focus outline

A focus indicator is whatever pixels change to show that a control is focused. The standard defines it that broadly on purpose. A swapped background colour, a thickened border and an inverted button all count, so the browser's own ring is one implementation rather than the whole category. Browsers draw something by default, and the decade-old habit of deleting it with outline: none and putting nothing back is still the commonest way a site becomes unusable by keyboard. The modern answer also removes the objection that started the habit. The :focus-visible selector shows an indicator to somebody navigating by keyboard and keeps it out of the way of a mouse click. That is what browsers now do with their own defaults. One more thing saves an argument later. Three separate rules meet on this one thing. A missing indicator fails 2.4.7. A visible one under 3:1 against what sits next to it fails 1.4.11, unless the browser drew it and you left it alone. A visible, contrasty one that covers too little area or barely changes between focused and unfocused is 2.4.13, and that one is Level AAA.

Focus indicator, in full →

On this siteRule 2.4.7 Focus Visible · Rule 1.4.11 Non-text Contrast

Focus order

Focus order is the sequence focus travels in as you press Tab, and by default it follows the order elements appear in the code rather than the order they appear on screen. Those two can be pulled apart completely by CSS, which is where the trouble starts. Flexbox and grid can move a block to the top of the page visually while leaving it last in the source, and a keyboard user then reaches it last. The rule is narrower than the folklore around it. What it asks for is an order that preserves meaning and operability, and W3C says plainly that focus order does not have to match the visual layout. More than one order can be right for the same content, so the test is never whether the order matches the picture. It is whether somebody moving through it can still understand the page and still finish the task.

Focus order, in full →

On this siteRule 2.4.3 Focus Order · Keyboard accessibility

Focus trap

Also called: Keyboard trap

A focus trap is somewhere on a page that keyboard focus can get into and not out of. A mouse user clicks past it and never knows it was there. Somebody on a keyboard presses Tab and watches it cycle round the same three controls, and their options narrow to leaving the page. It is a Level A failure, and it is one of the few that can end a session outright rather than just make it harder. Two things are worth knowing before you go looking for one. Keeping focus inside a modal dialog on purpose is not a trap, and it is the correct way to build one, as long as there is a way out. A dialog that is not modal is a different case, because nothing outside it has been switched off, so containing the tab sequence there needs a deliberate way back into the page. And a component that needs something other than Tab or the arrow keys to release focus still conforms if it tells the user which key does it. A media player that says press Escape to leave is passing, not failing.

Focus trap, in full →

On this siteRule 2.1.2 No Keyboard Trap · Modals and dialogs

Focus visible

Focus visible is two things with nearly the same name, and keeping them apart settles an argument. Rule 2.4.7 is the WCAG requirement, at Level AA, and its actual words are that a keyboard operable interface has a mode of operation where the focus indicator is visible. Not always visible. A mode. The CSS focus-visible pseudo-class is the other thing, a selector browsers match when their own heuristics decide an indicator would help, which in practice means the keyboard and other non-pointer input. That mode wording is what makes the pseudo-class safe rather than clever, because the keyboard is a mode where the indicator appears. And 2.4.7 sets no contrast ratio at all. The 3:1 people quote comes from rule 1.4.11, which excepts an indicator the browser drew and you never restyled, and the rules about the indicator's area and its change between states come from 2.4.13, which is Level AAA.

Focus visible, in full →

On this siteFocus indicator · Showing keyboard focus with focus-visible

Forced colors

Also called: forced-colors media query, forced-color-adjust

Forced colors is what the web platform calls it when somebody's operating system throws your palette away and repaints the page in colors they chose. It is the CSS side of high contrast mode, and it is the side you can write for, through a media feature named forced-colors and a set of system color keywords that carry the reader's own palette into your stylesheet. While it is on, the browser overrides your color on a named list of properties and does three more things nobody expects. Box shadows and text shadows compute to none. A background image computes to none unless it points at a real file, so gradients go and photographs stay. And the browser paints a solid plate behind text, which is what keeps the words on those photographs readable. The opt-out is a property called forced-color-adjust, and the specification is blunt about when to reach for it. Use it where you are adjusting the colors yourself to serve the same need, never because the mode made your brand look wrong.

Forced colors, in full →

On this siteHigh contrast mode, from the reader's side · Colour and contrast · Rule 1.4.11 Non-text Contrast · Testing themes and forced colors as states

Form label

A form label is the visible text naming a field, tied to that field in code so software can tell which words belong to which box. The visible half and the coded half are two different things and WCAG names them separately, which is worth knowing because a field can have one without the other. Placeholder text is the usual way that happens. It looks like a label, and it does get used as a last-resort name when nothing better exists, so a placeholder-only field is not silent. It is worse than silent in a different way. The prompt vanishes the moment somebody types, so anybody interrupted mid-form has lost the question they were answering, and placeholder grey is usually the palest text on the page. Four separate rules touch this. One asks that a label or instruction exists at all. One asks that the control carries a name software can read. One asks that the connection between them sits in the code. And one asks that the label describes the field.

Form label, in full →

On this siteRule 3.3.2 Labels or Instructions · Rule 4.1.2 Name, Role, Value · Rule 1.3.1 Info and Relationships

Terms starting with G

Top ↑

Guideline

A guideline is the middle layer of WCAG, and WCAG 2.2 has 13 of them sitting between the four principles and the rules anybody is ever tested against. Each one states a goal in a sentence, such as making it easier to see and hear content, or providing ways to help users navigate and find things. Nothing is ever measured against a guideline, which is the odd and useful part. Guidelines are binding content inside the standard and they are not testable, because what gets tested is the success criteria underneath them, alongside four conformance requirements that sit outside the criteria entirely. Once you know that, the numbering reads itself. In 1.4.3 the first digit is the principle, the second names the guideline, and the third is the criterion, so 1.4.3 is Perceivable, then Distinguishable, then Contrast (Minimum).

Guideline, in full →

On this siteThe full structure, mapped · WCAG versions and levels

Terms starting with H

Top ↑

Heading structure

Heading structure is the h1 to h6 outline of a page, and for screen reader users it is the map. Asked how they find something on a long page, 71.6% of people in WebAIM's survey said they navigate by heading. That is five times the share naming any other method. On a page with no real headings, the route is not there. The requirement is narrower than the advice, though, and it pays to keep them apart. WCAG asks that a heading be exposed as a heading rather than as text styled to look like one, and faking a heading with styling is the named failure. A real h1 to h6 element is the way almost everybody does it, and W3C also sanctions the heading role with a level attribute where a native element is out of reach. It sets no count for how many h1 elements a page may have and no flat rule against skipping a level, though a jump that misrepresents how the sections actually nest is a different argument and can be a finding on its own. One h1 naming the page, then h2 for sections and h3 inside those, is good HTML and it is our house style. It is not what anybody gets cited for.

Heading structure, in full →

On this siteRule 1.3.1 Info and Relationships · Rule 2.4.6 Headings and Labels · Check your heading outline

High contrast mode

Also called: Contrast themes, forced colors mode

High contrast mode is an operating-system setting that throws away most of a site's colours and paints the page in a palette the user chose. Windows ships the version the web can answer, now called Contrast themes. The web platform's name for what it does to your CSS is forced colors, and that entry covers the code side. On the reader's side the effect is short enough to list. The browser discards your author values for text colour, backgrounds, borders and outlines, and substitutes system colours. It forces box shadows and text shadows off. It drops any background image that is not a real image file, so gradients go and photographs stay. Text sitting on top of them stays readable because the browser draws a solid plate behind it. The failures are predictable once you know that list. Meaning carried in a gradient or a shadow disappears. A card drawn with background colours instead of real borders loses its edges. And a focus ring built from a box shadow stops existing for exactly the people most likely to have the mode turned on.

High contrast mode, in full →

On this siteForced colors, the CSS side · Rule 1.4.3 Contrast (Minimum) · Colour and contrast

Terms starting with I

Top ↑

Iframe

Also called: Inline frame, embed

An iframe embeds one document inside another. A map, a video player, a payment form, a chat widget. What matters is that it is a whole separate document with its own focus behaviour rather than a styled box, and every consequence follows from that. Each one carrying content a person can use needs a name saying what is inside it, and the conventional way to give it one is a title attribute on the iframe. That attribute sits on your markup rather than the vendor's, which is the answer to we cannot fix it, it is theirs. The W3C technique for it serves two rules. Giving a component a name, and grouping repeated content so it can be skipped.

Iframe, in full →

On this siteThird-party scripts and embeds · Give iframes descriptive titles

Inclusive design

Inclusive design means treating the full range of human variation as the starting point rather than as an edge case to accommodate once the design is finished. It is broader than accessibility in what it covers and looser in what it demands. Accessibility addresses the barriers that shut out people with disabilities. Inclusion takes in the rest of the spread too, and W3C's own list of what that means holds a few entries nobody expects. Alongside age and language it names hardware, literacy, economics, education and geography. The argument underneath is the one Microsoft states as solve for one and extend to many, which is the repeated finding that designing for a permanent disability ends up serving a far wider group. WCAG makes the same point about itself in passing. The rule asking for a visible focus indicator was written for keyboard users, and it names people with attention limitations, short-term memory limitations and limits in executive processes among those it helps.

Inclusive design, in full →

On this siteThe rules that encode it · Accessibility vs. usability · Where the three words differ

Informative

Informative means for information purposes and not required for conformance, which makes it the exact opposite of normative. It covers whole documents, the techniques and the Understanding pages among them, and it also covers six kinds of material inside WCAG itself. Introductory text, appendices, sections marked non-normative, diagrams, examples and notes are all informative, so a note sitting under a success criterion tells you how W3C reads that rule rather than adding to it. What the word does not mean is optional extra. W3C cautions against anyone requiring its own sufficient techniques, on the grounds that the only thing that should ever be required is meeting the success criteria.

Informative, in full →

On this siteSufficient, advisory and failure techniques · All 432 techniques

Terms starting with J

Top ↑

JAWS

Also called: Job Access With Speech

JAWS is the long-established commercial screen reader for Windows, made by Freedom Scientific, and the one named most often when people are asked which screen reader they use first. In WebAIM's most recent survey it led on that measure at 40.5%, with NVDA close behind at 37.7%, so it is the leader by a margin rather than the dominant product. Ask the question the other way and the order flips, because NVDA is the more commonly used of the two once people name everything they use. That is not a contradiction, it is two different questions, and most people answer both, since 71.6% of respondents use more than one screen reader. What settles it for testing is the pairing. JAWS with Chrome is the single most common screen reader and browser combination there is, at 24.7%, which is why it belongs in any test matrix that claims to be serious.

JAWS, in full →

On this siteHow we test with screen readers · How to test with a screen reader

Terms starting with K

Top ↑

Keyboard accessibility

Keyboard accessibility means every piece of functionality on the page can be operated through a keyboard interface. That is the actual wording of the rule and it is worth reading twice, because it is about functionality rather than about Tab reaching everything. Without a mouse is the everyday shorthand for it and the two are not the same, since a keyboard-driven mouse emulator does not satisfy the rule. Tab moves between controls, Enter and Space activate them, and arrow keys move inside a menu or a radio group. A menu that Tab enters once and arrow keys navigate through is the correct pattern, not a gap. The reason this serves so many people is a definition most readers never meet. What the standard asks for is operability through a keyboard interface, which is the software channel a program uses to receive keystrokes. Speech-to-text software, on-screen keyboards and switch devices all send keystrokes down it. So blind people, people with tremor or limited dexterity, and people who never touch a keyboard at all are served by one rule. Two things it does not cover. The rule carves out functions that depend on the path the user's movement takes rather than only where it starts and ends, such as free-hand drawing. And being fully keyboard operable does not on its own satisfy the rules about pointer gestures, because some people use a pointing device and cannot be precise with it.

Keyboard accessibility, in full →

On this siteRule 2.1.1 Keyboard · The keyboard accessibility guide

Keyboard shortcut

A keyboard shortcut is a second, faster route to something a page can already do, and the accessibility rule covers one narrow kind of it. Take a shortcut a page builds out of printable characters alone, meaning letters, numbers, punctuation or symbols, such as pressing s to open search. It has to be switchable off, remappable to something that includes a non-printable key, or live only while the relevant control has focus. The reason is speech input. Speech recognition software turns what somebody says into text and delivers that text as keystrokes. A person dictating into a field is emitting the very characters your shortcut is listening for. A stray s fires the command instead of typing a letter. Two things surprise people. Holding Shift to type the character does not get you out, because what the rule matches is the character rather than the keys that produced it, so a question mark shortcut is caught on any layout where the question mark needs Shift. And it covers sequences, so pressing g and then a in quick succession is caught by the same rule, which catches the teams who chose two-key sequences believing they were the safe option.

Keyboard shortcut, in full →

On this siteRule 2.1.4 Character Key Shortcuts · Keyboard accessibility

Terms starting with L

Top ↑

Landmark

Also called: ARIA landmark, page region

A landmark is a named region of a page that assistive technology can list and jump straight to, and there are eight of them. Banner, navigation, main, complementary, contentinfo, region, search and form. Modern HTML gives you most of them without any ARIA, with conditions worth knowing. The nav and main elements always produce one. The header and footer elements do it whenever they answer to the body, and they stop doing it once they sit inside an article, aside, nav, section or main. A plain wrapping div changes nothing, which is why the header inside a card built from divs is still announcing itself as the banner of your page. And section and form produce nothing at all unless you give them an accessible name. The one that earns its keep is main, because it is the sanctioned way to let somebody skip the navigation without a skip link, and fewer than half of home pages have one. Be careful how you sell the rest, though. Landmarks are the navigation people report using least, at 3.7% naming them as their first move on a long page against 71.6% for headings. They are a useful second mechanism rather than the main event.

Landmark, in full →

On this siteRule 1.3.1 Info and Relationships · Rule 2.4.1 Bypass Blocks

Level A, AA, AAA

Also called: WCAG conformance levels

Level A, AA and AAA are WCAG's three conformance levels, and they stack. Meeting AA means meeting everything at A as well, which is why nobody claims the two separately. At WCAG 2.2 that works out at 31 criteria at Level A, 24 at AA and 31 at AAA, so 86 in all and 55 in the set an audit aimed at AA tests against. Level A is the floor, and the criteria there tend to be the ones where assistive technology cannot make up for what is missing. Level AA is what the laws and the contracts name, near enough without exception. Level AAA is an enhanced set that W3C advises against requiring across a whole site, in a note attached to the conformance requirements. The reason it gives is plain. Some content cannot satisfy all of it. So when a law, a client or a lawsuit says WCAG compliant, they almost always mean Level AA. Reading it as anything else costs you money in one direction or credibility in the other.

Level A, AA, AAA, in full →

On this siteLevel A: the floor · Level AA: the legal standard

Live region

A live region is a part of the page marked so that assistive technology announces changes to it without the user having to go and look. A cart total updating, a save confirmation, a search count, a validation error appearing. Without one, all of that is silent, and somebody who cannot see the page clicks, hears nothing, and has no way to know whether it worked. The mechanical detail that decides whether yours works is the word changes. Assistive technology conveys changes to a live region rather than whatever was in it when the region first rendered, so the container has to exist on the page before the message arrives in it. Inject a fresh div with the message already inside and usually nothing is announced. The alert role is the deliberate exception, because ARIA asks the browser to raise a system alert event when one is created. Everything else is why so many developers decide live regions are unreliable, when what really happened is that the region and its content turned up together.

Live region, in full →

On this siteRule 4.1.3 Status Messages · ARIA explained · Announcements in a chat interface

Low vision

Low vision is sight loss that ordinary glasses cannot correct, and it covers far more ground than a single blurred-vision picture suggests. Reduced sharpness, a restricted field, blind spots, lost contrast sensitivity and light sensitivity all sit inside it, along with colour vision deficiency and the changes that come with age. It is much more common than total blindness, roughly six people in seven of everybody with a visual impairment. Most of it comes from conditions like cataracts, glaucoma and diabetes, which makes this largely an older audience. What catches designers out is that there is no single set of settings to design for. Needs inside this group genuinely conflict. Somebody with light sensitivity can find the high contrast another person needs unreadable, and somebody with tunnel vision and sharp sight may make text smaller to fit more words in view. That is why the answer is user control rather than one optimised design, and why the rules that serve this group are all about what survives being changed.

Low vision, in full →

On this siteRule 1.4.4 Resize Text · Rule 1.4.10 Reflow

Terms starting with M

Top ↑

Manual testing

Manual testing is a person working through a page against the rules a machine cannot decide. Unplugging the mouse and driving everything by keyboard. Listening to the page in a screen reader. Reading the alt text and asking whether it is true, reading the error messages and asking whether they help, and watching where focus goes when a dialog opens. It is not a supplement to a scan. It is the part of the standard scanning cannot reach, and W3C says as much in its own words, that evaluation tools cannot determine accessibility and can only assist in doing so. The strongest version of it adds testers who use assistive technology in daily life. Knowing how people with disabilities actually use the web is named as its own kind of expertise, separate from knowing the standard.

Manual testing, in full →

On this siteOur three-pass method · Why machines cannot do this part

Mechanism

A mechanism is a process or technique for achieving a result, and WCAG reaches for the word whenever it wants an outcome without prescribing a control. Seventeen success criteria use it. The mechanism does not have to be something you built, because the standard lets you rely on one the platform or the user agent already provides, which is why password manager support and copy and paste both count as ways of helping someone through a login. What the standard does not do is let the mechanism off. It has to meet every success criterion at the level you are claiming, so a pause button no keyboard can reach is not a mechanism for the purposes of the rule that asked for one.

Mechanism, in full →

On this siteRule 2.2.2 Pause, Stop, Hide · Rule 2.4.1 Bypass Blocks, the cleanest use of the word · Rule 1.4.4 Resize Text, where the browser is the mechanism

Motor disability

Also called: Mobility impairment

Motor disability covers anything that affects movement, strength, precision or sensation, from tremor and arthritis to multiple sclerosis, repetitive strain, paralysis and missing limbs. The web consequence is much the same whatever the cause, which is that mouse precision cannot be assumed. People navigate by keyboard, by switch, by voice, by eye tracking, with a head pointer or a mouth stick, or with a mouse they simply cannot aim well. The barriers overlap heavily across all of it. Small targets sitting close together. Anything that only happens on hover. A form that times out mid-answer. Anything that has to be dragged. The group is also much wider than a list of diagnoses. A broken wrist, a bad day with a fluctuating condition and holding a baby all produce the same functional need, and somebody in that position is far less likely to know an accessibility setting exists, let alone where to find it.

Motor disability, in full →

On this siteThe Operable rules · Rule 2.5.8 Target Size

Terms starting with N

Top ↑

Name, Role, Value

Name, role and value are the facts assistive technology needs about a control, and the shorthand quietly drops one. What the rule asks is that the name and the role can be read by software. Then that any state or value the user can change can also be set by software. Then that changes get announced to whatever is listening. The settable half is what makes a slider usable by voice control rather than merely readable, and it is the half nobody quotes. Native HTML gets you further than anything else and not all the way. Use a real button and the button role and its states come free. The name still has to come from somewhere. That is why an empty button element has a role and no name, and why an input with no label is announced by whatever the browser can scrape together. Build the same control out of a div and you start with none of it, and one piece has to be in place before another can work. A div's implicit role is one of the eleven ARIA forbids you to name, so a label on a roleless div is a label ARIA says not to write. The real role has to be there for the name to mean anything.

Name, Role, Value, in full →

On this siteRule 4.1.2 Name, Role, Value · ARIA explained

Narrator

Also called: Windows Narrator

Narrator is Microsoft's screen reader, and what matters about it is the shape of its use rather than its feature list. It is built into Windows with nothing to download, and in WebAIM's most recent survey it is the main screen reader of 0.7% of the people who answered while 37.3% of them use it. That gap is the whole story. Hardly anybody picks it, and more than a third of screen reader users have it in rotation, because it is the one already sitting on a machine that is not theirs, and the one that gets a computer talking again when the paid reader will not start. Its reading state is called scan mode and it switches itself on in Edge, Chrome and Firefox. It appears in none of the common screen reader and browser pairings that same survey publishes, which is the honest answer to whether it belongs in your test matrix.

Narrator, in full →

On this siteHow to test with a screen reader · How our three passes work

Non-interference

Non-interference is the fifth conformance requirement, and the four rules people quote from it are only its second half. The first half is about the technologies you are not relying on to meet the criteria. Those must not block access to the rest of the page, which means the page has to keep conforming while such a technology is turned on, while it is turned off, and where a user agent does not support it. Four success criteria are then added on top, and they apply to all content on the page, including content not otherwise relied upon to meet conformance. Audio that starts by itself, runs past three seconds and gives you no way to stop it or turn it down on its own. Keyboard traps. Flashing above the threshold. And moving, blinking, scrolling or auto-updating content that starts by itself and offers no way to pause, stop or hide it. None of it can be set aside, because no part of a page can be set aside from conformance at all.

Non-interference, in full →

On this siteRule 1.4.2 Audio Control · Rule 2.1.2 No Keyboard Trap · Rule 2.3.1 Three Flashes · Rule 2.2.2 Pause, Stop, Hide

Normative

Normative means required for conformance. The main content of WCAG is normative, and the parts anything ever gets tested against are the success criteria and the five conformance requirements alongside them. The four principles and thirteen guidelines above them set the objectives, and the guidelines are not testable at all, so nothing is ever measured against one. The line also runs inside the standard rather than only around it. Introductory material, appendices, sections marked non-normative, diagrams, examples and notes are all informative, so a note sitting under a success criterion tells you how to read the rule rather than adding to it. Techniques and Understanding pages sit outside WCAG altogether, and they are informative too.

Normative, in full →

On this siteThe conformance requirements · Sufficient, advisory and failure techniques

NVDA

Also called: NonVisual Desktop Access

NVDA is a free, open-source screen reader for Windows, made by NV Access, and more people reach for it than for anything else. That comes from WebAIM's survey, which is the closest thing this field has to a headcount. On the commonly-used measure NVDA leads at 65.6%, and on the narrower question of which one people name as their primary it sits second at 37.7%, just behind JAWS. Cost is not the headline reason people give, which is the thing most write-ups get wrong. Asked why they use their main screen reader, respondents put existing comfort and expertise first at 45.8%, features second, availability third, and cost fourth at 9.2%. That question covered every screen reader at once rather than this one, so read it as what moves the field. What being free changes is who gets to use a screen reader at all. NVDA also carries the highest satisfaction of any measured, at 97.6%. And the geography inverts. It leads JAWS in Europe, and across Asia and across Africa and the Middle East it runs at around 70% primary usage.

NVDA, in full →

On this siteHow we test with screen readers · How to test with a screen reader

Terms starting with O

Top ↑

Operable

Operable is the second WCAG principle, and its one-line statement is that user interface components and navigation must be operable. In practice that means everything works by keyboard and nothing traps focus. People get enough time. Nothing flashes at a rate that can trigger a seizure. Navigation helps you find things and know where you are, and targets are big enough to hit. It is the largest principle in the standard, at 34 criteria across five guidelines, and it is where WCAG 2.2 did most of its work, adding five of its nine new criteria. If Perceivable is about taking information in, Operable is about being able to act on it, and the difference shows in the consequences. A missing alt attribute costs somebody information. A keyboard trap costs them the session.

Operable, in full →

On this siteAll Operable rules · Keyboard accessibility

Overlay

Also called: Accessibility widget, overlay widget

An overlay is a third-party script you add to a site, sold on a subscription, that shows an accessibility toolbar and tries to repair the page automatically as it loads. It does repair some things. The field's own reference document on overlays says a non-trivial array of accessibility problems can be fixed this way, and pretending otherwise is both wrong and easy to disprove. What it cannot do is repair enough of them to reach conformance, and the categories it handles worst are the ones that decide whether a site is usable at all. Text alternatives, form labels and error handling, and keyboard access. The record around the marketing is separate and harder. The Federal Trade Commission approved a final consent order in April 2025 requiring one major vendor to pay $1 million that may be used to provide refunds to consumers. That settled a deception case. The order also bars the vendor from claiming its automated products can make any website conform, or keep it conforming, without evidence to back that up. Half of that order is about something else entirely, which is disguised reviews presented as independent opinion.

Overlay, in full →

On this siteOverlays: the documented record · The vendors, compared

Terms starting with P

Top ↑

Partial conformance

A statement of partial conformance is what you publish when a page fails only because of content outside your control, such as user comments, an ad slot, or a third-party widget you cannot repair. It is a statement of non-conformance rather than a conformance claim with a carve-out, and the sentence you are given to write opens "This page does not conform". What follows is the level the page would reach if the named parts were removed. WCAG has two of these statements. One covers third-party content, and the other covers a page that would conform if accessibility support existed for the language it is written in. Neither one covers content you chose to embed and could replace.

Partial conformance, in full →

On this siteThird-party scripts · Writing an accessibility statement · The conformance requirements

PDF/UA

Also called: ISO 14289-1, ISO 14289-2

PDF/UA is the international standard for accessible PDF, published as ISO 14289 in two parts. Part 1 covers files built on PDF 1.7 and part 2 covers PDF 2.0. They are siblings rather than successive versions, so which one applies depends on the version of PDF you are producing. It describes the finished file rather than the way you make it. What has to be tagged, how the tag tree is structured, what metadata has to be there. It says nothing about how to produce one. Its custodian is clear about something else, too. Conforming to it does not by itself make a document accessible. It has a colour checkpoint and it asks only that meaning carried by colour also lives in the tags, so there is no contrast ratio to measure anywhere in it. It has scripting checks and they cover particular behaviours rather than whether your interactive form can be used. And whether the words can be understood is outside it entirely. Those are the parts WCAG covers, which is why the two are companions and why laws name WCAG.

PDF/UA, in full →

On this sitePDF accessibility, explained · Check a PDF free

Perceivable

Perceivable is the first of WCAG's four principles, and the standard states it as information and user interface components being presentable in ways people can perceive. The second half of that sentence is the half that gets dropped, and it is why the contrast rule for buttons and icons lives here alongside the one for text. In practice it means images carry text alternatives, and audio and video carry captions and descriptions. Structure sits in the markup rather than only in the styling. And nothing depends on colour alone or falls apart when somebody enlarges the text. It holds 29 criteria across four guidelines, spread unevenly in a way worth knowing. Text alternatives is a single criterion carrying that whole apparatus, while the guideline about telling content apart from its background holds 13, seven of them at Level AA. That is where most of the AA work on a normal site turns out to be.

Perceivable, in full →

On this siteAll Perceivable rules · How to write alt text

Plain language

Plain language is writing that lets somebody find what they need, understand it the first time, and act on it. That is the international standard's own definition, and the important part of it is what it measures. It is about how successfully a reader can use the document rather than about any mechanical property of the text. A readability formula is a tool for that and never the test itself, which is roughly where W3C leaves it. Its current pages name no particular formula and its technique for this simply says a readability formula is applied. WCAG's own reading-level rule sits at Level AAA, so most sites carry no WCAG obligation for their body copy, and it shows. Other rules can still reach it, and so can a plain-writing duty attached to a sector or a contract, which is worth checking before you assume nothing applies. This is the first thing teams drop, and the accessibility work that needs attention more than it needs budget. It is not entirely optional under WCAG either. The Level A rules asking for labels, instructions and error messages a person can act on cover exactly the text where clear words matter most, which is also where W3C says to start.

Plain language, in full →

On this siteThe Understandable rules · Cognitive accessibility

Pointer gesture

A pointer gesture is a touch or mouse action where the path your finger takes matters, or where more than one finger is on the screen at once. Swiping, tracing a shape, pinching to zoom, a two-finger rotate. Rule 2.5.1 asks at Level A that the same job can be done with a single pointer and no path, and W3C names six ways of doing it. A tap, a click, a double tap, a double click, a long press, or a click and hold. So the fix is usually smaller than people fear. One finger is not an escape either, because a single-finger swipe is a single pointer interaction and a path-based gesture at the same time. What the rule does not reach is the software around your content. Gestures the operating system, the browser or a screen reader needs are not yours to replace.

Pointer gesture, in full →

On this siteRule 2.5.1 Pointer Gestures · Mobile accessibility · Pointer routes in a canvas editor

POUR

POUR is the field's shorthand for WCAG's four principles, which are Perceivable, Operable, Understandable and Robust. Every guideline and every rule in the standard sits under exactly one of them, so it is the fastest way to hold the whole thing in your head. Can somebody take the information in, can they work the controls, can they follow what is happening, and will it keep working with the software they use? Two things about the acronym are worth knowing. It appears nowhere in the standard itself, which always names the four in full, so it is a teaching device rather than something you can point at in the specification. And the four are nothing like equal. Perceivable holds 29 criteria and Operable 34, Understandable has 21, and Robust has two.

POUR, in full →

On this sitePerceivable · Operable · Understandable · Robust

prefers-reduced-motion

prefers-reduced-motion is a preference people set in their operating system or their browser, and it reaches your CSS as a media query and your JavaScript as a check you can read. Set to reduce, it says the person wants motion-based animation removed or replaced. Two groups ask for it and most write-ups name only one. People with vestibular motion sensitivity, for whom large movement brings on dizziness, nausea, migraine and sometimes a day in bed to recover. And people with attention deficits, for whom the same movement is a distraction they cannot work through. The criterion behind it is rule 2.3.3, which covers animation triggered by something the person did and sits at Level AAA. Motion that starts by itself answers to rule 2.2.2, and that one is Level A, so it is the one inside every legal obligation there is. The two are not a clean split, and W3C says as much. One animation can fail both.

prefers-reduced-motion, in full →

On this siteMotion and animation · Honouring prefers-reduced-motion in CSS

Process

A process is a series of user actions where every action is needed to finish the job, such as a checkout, a booking, or an application form. Conformance then works differently from anywhere else in the standard. Every page in the process has to conform at the level you are claiming. If one of them does not, none of them does. That includes the pages that would pass perfectly well on their own. That rule holds at A, AA and AAA alike. It is about actions rather than addresses, too, so a registration form sitting behind a single test counts as a process even when there is only one URL involved.

Process, in full →

On this siteThe conformance requirements · Rule 2.4.5 Multiple Ways, which excepts a step in a process · Rule 3.3.7 Redundant Entry, judged within one process · A checkout as a complete process · A job application across two systems

Programmatically determined

Programmatically determined is WCAG's term for information software can read out of the code rather than infer from how the page looks. It is a defined term, and there is more inside it than the usual paraphrase. Three conditions are packed in there. The data has to come from you, the author, rather than be guessed at by something clever. More than one kind of software has to be able to extract it, not just the one screen reader you happened to test with. And the point of extracting it is presenting the same information in another form, which is what makes this an accessibility requirement rather than a machine-readability one. A heading that looks like a heading because it is large and bold is determined visually. A heading marked up as an h2 is determined programmatically, because the element says what it is and anything reading the page can act on that. There is a second, far less quoted half of the same idea. Programmatically set is the write direction, and it is what lets voice control or a screen reader's own controls change a value rather than only announce it.

Programmatically determined, in full →

On this siteRule 1.3.1 Info and Relationships · ARIA explained

Terms starting with R

Top ↑

Reading level

Reading level is how much schooling a piece of text assumes, and WCAG measures it in years rather than in ages. The standard defines lower secondary education as ending nine years after primary school begins. Rule 3.1.5 asks that where text needs more reading ability than that, you also offer supplemental content, or a version that does not need more reading ability than that. Two things make it less alarming than it sounds. Proper names and titles come out of the text before anybody assesses it, and that exclusion is written into the rule itself rather than into a note. And it sits at Level AAA, so most sites have no formal obligation here. There is also a third way to satisfy it that the wording hides. Write the original clearly enough and no alternative is needed at all.

Reading level, in full →

On this sitePlain language · Cognitive accessibility

Reading order

Reading order is the sequence content reaches assistive technology in, and it comes from the source rather than from the layout. CSS grid and flexbox can move things around the screen without touching the source, which is what they were designed to do. Both specifications also say in as many words that the reordering is for the visual rendering only. It must not stand in for getting the source order right. CSS is starting to offer a deliberate way to change the order after all, and one browser has shipped it, so treat that as a thing to watch and keep the source order sound in the meantime. WCAG asks less than that. Rule 1.3.2 bites only where the sequence carries meaning, and W3C is explicit that more than one order can be correct and only one of them has to be provided. So the main content and the navigation in either order is fine. Reordering the rows of a table usually is not, because the order of rows is normally carrying something.

Reading order, in full →

On this siteRule 1.3.2 Meaningful Sequence · Tab order

Real-time text

Also called: RTT

Real-time text is a conversation in text where what you type reaches the other person as you type it, rather than when you press enter. The standards define it by how it feels rather than by the wire. The European Accessibility Act's own words are that the text is sent so the communication is perceived as continuous on a character-by-character basis. The European technical standard puts a number on that. Half a second from a composed unit of text being ready to it leaving for the network. It is what lets a text conversation run at speaking pace, with interruption and turn-taking, for people who are deaf or have a speech disability. It is the replacement for TTY, and WCAG has nothing to say about it at all.

Real-time text, in full →

On this siteEN 301 549 · The European Accessibility Act

Reflow

Reflow is content rearranging itself to fit a narrow viewport instead of making you scroll sideways to read every line. The rule behind it is written as a viewport size rather than as a zoom level. What it asks for is a width equivalent to 320 CSS pixels, with no scrolling in two directions at once and nothing lost on the way. Content that disappears at that width fails it just as surely as content you have to chase sideways. The familiar 400% figure is only what that width comes to if you start from a window 1280 CSS pixels wide, which is why narrowing a window is a cleaner test than zooming. It does not require a single column, so a carousel or a side-by-side comparison can pass. And parts of a page that genuinely need two dimensions are excused outright, which covers maps, diagrams, video, games, presentations and data tables.

Reflow, in full →

On this siteRule 1.4.10 Reflow · Mobile accessibility

Relied upon

A technology is relied upon when the content would stop conforming if that technology were turned off, or if a user agent did not support it. Both halves of that test matter, because a page can survive one state and fall over in the other. Turn the script off and the tab strip you built out of it is gone. Meet a user agent that never supported it and the same thing happens for a different reason. What gets tested is a web content technology, meaning the formats and languages you author in, so HTML, CSS, JavaScript, SVG and PDF all qualify. The software people open them with does not, so a browser, a media player or a screen reader is never the thing being relied upon. The question is never whether you used something. It is whether the page still works without it.

Relied upon, in full →

On this siteThe conformance requirements · User agent, the category this one is not · Third-party scripts

Remediation

Remediation is the work of fixing the accessibility problems an audit found, in the code, the content and the design, until the failures are gone. An audit tells you what is broken and why. Remediation is what makes it work, and the two are usually bought separately because they are two different jobs. One thing to know before you plan it. Conformance is judged a whole page at a time. Nine fixes out of ten leaves that page exactly where it started as far as a claim goes, if the tenth is a failure at the level you are claiming, however much better the page is to use. The order that works is by user impact first, in your own code rather than under a widget layered on top, and then retested.

Remediation, in full →

On this siteEvery fix, explained · What an audit costs

Robust

Robust is the fourth of WCAG's four principles, and it asks that content be built well enough for a wide variety of user agents, including assistive technology, to interpret it. Interpret, not render. A page can paint perfectly and still tell a screen reader nothing about what any of it is. It is the smallest principle in the standard by a distance, one guideline and two live rules against Operable's thirty-four, and it is the only one that adds nothing at all at Level AAA. It is also where hand-built controls fail hardest, because those two rules ask for exactly the things a div forgets. A name, a role and a value, and an announcement when a status changes.

Robust, in full →

On this siteAll Robust rules · ARIA explained

Role

A role is what kind of thing an element is, as reported to assistive technology. Button, link, heading, checkbox, navigation. The standard's own wording is that a role identifies the element's function, which is a better way to think about it, because putting role="button" on a <div> does not turn it into a button. It tells assistive technology what the author intends it to do. Native HTML carries one for free, so a <button> element is announced as a button with nobody declaring anything. A clickable <div> is the case people ask about, and it is not quite roleless. Its role is generic, which means a nameless container, and ARIA forbids you to name one of those, so an aria-label on a bare <div> is worse than useless. Give it a role first. Then the name sticks.

Role, in full →

On this siteRule 4.1.2 Name, Role, Value · ARIA explained

Terms starting with S

Top ↑

Screen magnifier

Also called: Zoom software

A screen magnifier enlarges part of the screen, from the zoom built into every operating system up to dedicated software like ZoomText. That one runs from 1x to 60x, with fractional steps such as 1.4x at the bottom of the range, and down there a user still sees most of the page. Push the magnification up and the view narrows to a small moving window, so two things far apart on screen are never seen together. A confirmation drawn on the opposite side of the page from the button just clicked may as well not have rendered. Magnification is not browser zoom. Magnifying the screen enlarges what is already drawn and leaves the viewport alone, so the page does not reflow to fit, which means the reflow rule that rescues a zoom user does nothing in that mode. The rules about where things sit relative to each other still do a great deal.

Screen magnifier, in full →

On this siteRule 1.4.13 Content on Hover or Focus · Assistive technology, in full

Screen reader

A screen reader is software that reads the screen out loud, or sends it to a braille display, for people who cannot see it. It is driven by keyboard or by touch gestures, jumping between headings, links and form fields and announcing each one. "Heading level 2, Pricing." "Link, Contact us." "Edit, Email address." On a desktop, JAWS and NVDA are each the main screen reader for about four survey respondents in ten, and VoiceOver for about one in ten. The picture changes completely on a phone. Nine in ten of the same people also use a screen reader there, and VoiceOver is overwhelmingly the one they reach for, so testing that stops at a laptop has skipped a mode most of the audience uses as much as the one you tested. What none of them can do reliably is announce something the code never exposed, and that is why a site can look flawless and be unusable.

Screen reader, in full →

On this siteHow our blind testers work · JAWS · NVDA · VoiceOver

Section 508

Section 508 is the US law requiring federal departments and agencies, plus the Postal Service, to make their digital products accessible. Of its own force it reaches nobody else. It does not follow federal funding, which is Section 504's job, and it does not bind suppliers directly either. It reaches them through purchasing, so if you want government contracts, procurement will ask for proof, usually a completed VPAT. The thing worth knowing about the standard itself is its age. Section 508 has incorporated WCAG 2.0 Level A and AA since January 2018 and still does, while the ADA Title II rule and the health department's Section 504 rule both name WCAG 2.1 AA. A vendor selling to a federal agency and a state agency is testing against two versions of the same standard for two arms of the same government.

Section 508, in full →

On this siteSection 508, explained · VPAT · Section 255, the telecoms regime next door

Semantic HTML

Semantic HTML means using HTML elements for what they mean rather than for how they look. A <button> for something that acts, <nav> for navigation, <h2> for a heading, <table> for tabular data. Those semantics are what assistive technology reads. Where the element is a control, the native one arrives with four things already correct. A role, keyboard behavior, focus participation, and state reported to the platform. Structural elements bring the first of those and none of the rest, because a heading was never meant to be operated. Rebuild a control out of <div>s and you owe all four by hand. An ARIA role changes what the accessibility tree says the element is and changes nothing else, so none of the work the browser was doing for you comes back with it. That is why so many expensive ARIA repairs turn out to be undoing a decision made in the markup.

Semantic HTML, in full →

On this siteRule 1.3.1 Info and Relationships · The HTML techniques, explained

Set of web pages

A set of web pages is a collection that shares a common purpose and is created by the same author, group or organisation. Both halves have to hold, which draws the boundary in places people do not expect. The standard's own examples do the drawing. A checkout that strips the site navigation is a different set from the product pages it came from. A blog on a subdomain with its own navigation, written by a different group of people, is a different set again. And a translation is always separate from the original. Five success criteria are judged across a set rather than page by page, which is why consistency gets checked by opening several pages side by side instead of reading one of them carefully.

Set of web pages, in full →

On this siteRule 3.2.6 Consistent Help · Rule 3.2.3 Consistent Navigation · Rule 3.2.4 Consistent Identification · Rule 2.4.5 Multiple Ways · Rule 2.4.8 Location · Scoping a configurable product

Sign language interpretation

Sign language interpretation is a signed version of spoken content, usually a video track running alongside the main one. WCAG asks for it in one place, rule 1.2.6, at Level AAA, and only for prerecorded audio inside synchronized media. So a podcast with no video sits outside it, and so does anything live, because no criterion at any level asks for signed live content. It matters because sign languages are languages. The standard's own note says true sign languages are independent languages, unrelated to the spoken language of the same country or region. Which is why a signed track serves one audience rather than all deaf viewers, and why written captions are a second language for some of the people watching.

Sign language interpretation, in full →

On this siteVideo, captions and transcripts · Level AA vs. AAA

Status message

Also called: status update

A status message is a change in content that tells you how something went, without moving focus to it. The definition names four kinds. The success or result of an action, a waiting state, progress through a process, and the existence of errors. Spinners and progress bars are in scope, which surprises people, because the confirmations get all the attention. The defining feature is the one people miss. If focus moves to the message it is not a status message and the rule does not apply, since a change of context has already interrupted the reader and been announced anyway. The rule exists for everything else, the text that appears quietly somewhere on the page while a sighted reader catches it out of the corner of an eye.

Status message, in full →

On this site4.1.3 Status Messages · Live region · Notifications and toasts

Success criterion

A success criterion is a single testable rule in WCAG, and it is the unit everything else is built from. Each one has a number like 1.4.3, a name like Contrast (Minimum), a level of A, AA or AAA, and conditions you either meet or do not. WCAG 2.2 has 55 of them at Levels A and AA and 86 across all three levels, which is why you will see both figures quoted and both are right. The A and AA ones are what laws, contracts and purchasing questionnaires almost always name, and we explain every one of them. The layer above, four principles and thirteen guidelines, is not testable at all, so nothing is ever measured against a guideline and "we meet Perceivable" means nothing.

Success criterion, in full →

On this siteAll 55 rules, explained · All 55 rules, listed

Switch device

Also called: Switch access

A switch device is an input for people with severely limited movement, and the picture most people have is a single large button pressed with a hand, head or knee. That picture is now only half of it. On a current phone a switch can be a mouth sound, a head turn the front camera picks up, or a pinch of finger and thumb. It can be a double tap on the back of the case. None of those needs extra hardware. Software then steps through the interactive elements on screen and the user fires the switch when the highlight reaches the one they want. On the web that stepping runs on whatever the keyboard interface can reach, which is why keyboard support is the work that serves these readers. Anything mouse-only drops out of the scan and is left to a slower fallback that aims a pointer by hand.

Switch device, in full →

On this siteRule 2.1.1 Keyboard · Assistive technology, in full

Terms starting with T

Top ↑

Tab order

Tab order is the sequence keyboard focus moves through when you press Tab, and it is field vocabulary rather than the standard's word. WCAG calls it focus order and HTML calls it sequential focus navigation order. It follows the source, and the thing that changes it is a positive tabindex, which pulls an element ahead of everything that has not been given one. CSS does something different and it is by far the more common trouble in current code. Flexbox ordering and grid placement move things on screen without moving them in the source, so the tab order stays exactly where it was and no longer matches what the reader is looking at. It is close to reading order and not the same thing, since reading order covers everything announced and tab order covers only what can take focus. Neither rule demands a particular order. Both bite only where the sequence affects meaning or operation.

Tab order, in full →

On this siteFocus order · Tabindex

Tabindex

Tabindex is an HTML attribute that decides whether an element can take keyboard focus and where it sits in the tab order. A value of 0 puts an element into the natural order, which follows the source. A value of -1 makes it focusable by script, and by a mouse click, while skipping it when tabbing, which is how you move focus to a heading or to an error message. Anything positive is the documented anti-pattern. Positive values go first, ahead of everything set to 0 or to nothing at all, and they run lowest number first, so a single stray tabindex="3" reorders everything around it. Every element added afterwards then has to be reasoned about against a custom order somebody left behind. Treat them as a bug, not a tool.

Tabindex, in full →

On this siteRule 2.4.3 Focus Order · Keyboard accessibility

Tagged PDF

A tagged PDF carries an invisible structure layer naming each part of the document. This is a heading, this is a list, this is a table header, this is decoration. The layer is stored separately from the marks on the page, with pointers between the two. Which is why the order of the tags can be right while the order the ink was painted in is not. Tagging is optional in the PDF standard itself. A file with no tags is a completely valid PDF, and the flag recording that a document follows the tagging conventions defaults to false. So a tool that emits an untagged file has broken no PDF rule. That is the whole reason accessibility has to be added on purpose. Where anybody has measured, the picture is poor. A PDF tooling vendor ran an automated check across nearly 70,000 PDFs on German public sector websites in 2026, and 37.4% carried no tags at all while fewer than one in ten passed.

Tagged PDF, in full →

On this siteHow to tag a PDF · Check whether yours is tagged

Target size

Also called: Touch target

Target size is how big a clickable or tappable thing is, and two WCAG rules govern it at two different numbers. Level AA asks for 24 by 24 CSS pixels, with a spacing route out for anything smaller. Level AAA asks for 44 by 44 and gives you no spacing route at all, which is where the familiar 44 actually comes from. The platform vendors then publish their own figures, and they do not agree with the standard or with each other. Apple's guidance says 44 points and Google's says 48dp, and neither of those is what a WCAG audit measures you against. One more thing that catches people. A small target on its own does not fail the AA rule. Small targets packed together do, which is why icon rows, close buttons, table row actions and pagination are where this bites.

Target size, in full →

On this siteRule 2.5.8 Target Size (Minimum) · Mobile accessibility

Technique

A technique is one of the W3C's documented ways of meeting, or of failing, a WCAG rule. Each one carries an ID whose letter says which technology it is for, so H is HTML, C is CSS, F is a documented failure, and G is general, meaning it holds whatever you built the page in rather than pointing at particular code. They come in three kinds. A sufficient technique is a documented way of meeting a criterion, though not always alone, because W3C joins some of them with AND and then the whole set is the route. An advisory technique goes past what a criterion requires, or covers only part of one, or uses something support has not caught up with. And a failure describes a particular way of getting it wrong. W3C publishes 432 of them for WCAG 2.2 across all three levels, and we have explained every one. The thing to hold on to is that all three kinds are informative. Content can satisfy WCAG without using a single documented technique, so an auditor who fails you for not using a named one has confused the advice with the rule.

Technique, in full →

On this siteAll 432 techniques, explained · Sufficient vs. advisory vs. failure

Text alternative

Also called: text equivalent

A text alternative is the standard's own term for text attached to something that is not text, in a way software can follow from one to the other. It can be the words themselves, the way alt text sits on a photograph. It can also be a short string pointing at a longer description elsewhere. A chart carries a few words saying that a description follows, and the description itself is written out in the paragraph underneath. What the alternative has to achieve comes from the rule rather than from the definition, and the rule asks for text that serves the equivalent purpose. That is the test to apply. Not what the thing looks like, but what it was there to give the reader.

Text alternative, in full →

On this siteAlt text · 1.1.1 Non-text Content · How to write alt text

Text to speech

Also called: TTS

Text to speech is the conversion of written text into a synthesized voice. It is one output of a screen reader and it is not a screen reader, which is the distinction that settles most arguments about it. W3C defines the two separately. Text to speech turns text into a voice reading it aloud. A screen reader processes the content, converts it to speech and to braille, and adds the shortcut keys, the reading modes and the structural navigation on top of that. Used on its own, in a browser read-aloud feature or a reading app, it speaks the text that application decides to hand it, which is normally the visible text rather than the structure around it. So it asks less of your markup than a screen reader does. What it always needs is text from somewhere. A picture of words gives it nothing until something else has pulled the words out first.

Text to speech, in full →

On this siteScreen reader · Assistive technology

Timeout

Also called: Session timeout, time limit

A timeout is a time limit the content sets, such as a session that expires, a booking held for ten minutes, or a form that clears itself after a while. The rule that governs it gives you six ways out, and most people only know three. Let the user turn the limit off before they meet it. Let them adjust it before they meet it, over a range of at least ten times the default. Or warn them before it expires and give them at least twenty seconds to extend it with one simple action, at least ten times over. The other three are exceptions, and they are narrow. A real-time event such as an auction where no alternative limit is possible, a limit that is essential because extending it would invalidate the activity, and anything longer than twenty hours, which this rule then leaves alone.

Timeout, in full →

On this siteRule 2.2.1 Timing Adjustable · Time limits and timeouts · Booking holds, and the two ticket examples

Toast notification

Also called: Snackbar, flash message

A toast notification is a short message that appears, usually to confirm something worked, and takes itself away on a timer. The timer is what people argue about, and the argument is normally aimed at the wrong thing. How long it stays is not the test. W3C's own worked example is a five-second toast that conforms, because the same information sits elsewhere on the page. Where it does not, that timer becomes a time limit the person has to be able to control. Three rules meet here. Whether it is announced at all, which is rule 4.1.3. Whether the timing can be adjusted, which is 2.2.1 at Level A. And whether it should be on a timer at all, which is 2.2.3 at Level AAA, where W3C's advice is to avoid alerts that disappear by themselves.

Toast notification, in full →

On this siteLive region · Rule 4.1.3 Status Messages

Tooltip

A tooltip is a small label that appears when something takes hover or focus. Rule 1.4.13 governs how it behaves once it is there, at Level AA, and sets three conditions. It has to be dismissible without moving the pointer or the focus, usually with Escape, unless it reports an input error or covers nothing. It has to stay put while the pointer travels onto it. And it has to remain visible until one of three things happens. The trigger is removed, the person dismisses it, or its information goes out of date. So a tooltip closing when the mouse moves away is expected behaviour rather than a failure. What this rule does not do is require a hover tooltip to answer to the keyboard as well. That obligation comes from rule 2.1.1, and it is why hover-only tooltips fail.

Tooltip, in full →

On this siteRule 1.4.13 Content on Hover or Focus · The hover content technique

Transcript

A transcript is the text version of audio or video, and two different things go by that one word. A basic transcript carries the speech and the non-speech sound you need to follow the content. A descriptive transcript adds the visual information as well. That second kind is what somebody who is both deaf and blind actually needs, because they reach your video as text on a braille display. The distinction decides what you have satisfied. For a podcast or any audio-only recording, a transcript is the Level A answer, and the only content excused is a recording that is itself a labelled alternative to text already on the page. For video with sound, a descriptive transcript is one of the two ways to meet the Level A description rule, the other being audio description itself. It stops being a route at Level AA, where audio description is the only answer, and it never replaces captions.

Transcript, in full →

On this siteRule 1.2.1 Audio-only and Video-only · Video, captions and transcripts

Terms starting with U

Top ↑

Understandable

Understandable is the third of WCAG's four principles, and it splits in two the way its own sentence does. The information has to make sense, and the operation of the interface has to make sense. Three guidelines sit underneath it. Readable covers declaring the language of the page, and of foreign-language passages inside it that are not proper names, technical terms or words the surrounding text has absorbed. Predictable covers navigation staying consistent, and the several separately scoped rules about the page not changing under somebody without warning. Input assistance covers labels, error messages that explain themselves, and help getting a form right. Those three names are what let you place any rule whose number starts with a 3. And nothing conforms to a principle, so "we meet Understandable" says nothing on its own. The criteria underneath are what gets tested.

Understandable, in full →

On this siteAll Understandable rules · Cognitive accessibility

Universal design

Universal design is the idea that one design should work for as many people as possible without anybody needing a special version or a separate adaptation. It started in architecture, where the example everyone reaches for is the curb cut, built for wheelchairs and used by anyone with a stroller, a suitcase or a delivery trolley. It is a treaty definition now as well, and the treaty's version covers programs and services alongside products and environments. That is what brings a website, a booking flow and a support line inside the term at all. The treaty also adds a sentence people miss. Universal design does not mean assistive devices stop being needed, and it was never meant to. Norway named the regulation behind its web accessibility duty after it, and built the phrase into the statute's own section headings.

Universal design, in full →

On this siteNorway's universal design law · Where universal design came from · Accessible web design

User agent

A user agent is any software that retrieves and presents web content, which takes in browsers, media players, plug-ins, and assistive technology as well. Assistive technology arrives by either of two routes. It can act as a user agent in its own right, the way a self-voicing browser does, or it can work alongside a mainstream one, the way a screen reader usually works with Chrome. The category matters because WCAG splits the duties. Your job is the content and the user agent's job is presenting it, which is why several rules are met by not interfering rather than by adding anything at all. Zoom is the clearest case. The browser already does it, and your responsibility is not to prevent it.

User agent, in full →

On this siteRule 1.4.4 Resize Text · Assistive technology

User testing with disabled users

User testing with disabled users means watching real people with disabilities try real tasks on your site. Not a developer imitating a screen reader, but somebody who uses one every day, working through your checkout. Watching them finds usability problems that a conformance check on its own does not, and that is the whole case for doing it. The reverse holds just as firmly, and it is the half most people selling this leave out. A set of sessions cannot tell you whether a site is accessible, because no realistic number of them covers the range of disabilities, strategies and tools a standard exists to cover. That is what the standard is for. So it is the third pass of our audits, not a substitute for the first two.

User testing with disabled users, in full →

On this siteHow our testing sessions work · Testing with blind users

Terms starting with V

Top ↑

Vestibular disorder

A vestibular disorder is a condition affecting balance and spatial orientation, and it runs through the inner ear and the brain rather than the inner ear alone. That matters, because some of the conditions most relevant to screens are central rather than inner-ear ones. Motion on screen can trigger real dizziness, nausea and migraine, and the harm is physical and immediate rather than a matter of difficulty. Large movement is the trigger. Parallax scrolling, full-page transitions, backgrounds that scale, anything that moves a big share of the view at once. The order we would take it in has three steps and only the last two are things W3C documents as ways of passing. Leave the large movements out, because a page with no unnecessary motion has nothing to disable. Offer your own control for the ones that stay. And honor the reduce-motion setting, which every mainstream desktop and mobile platform now carries somewhere in its settings, in your stylesheet and in your scripts alike.

Vestibular disorder, in full →

On this siteRule 2.2.2 Pause, Stop, Hide · Motion and animation

Voice control

Also called: Speech input, Dragon, Voice Access

Voice control means operating a computer by speaking, saying things like "click Submit" or "scroll down", and it is built into macOS, iOS, Windows and Android. It works by matching what you say against the labels you can see, which is why WCAG requires a control's accessible name to contain the text printed on it. Note that word. Contain, not match. A button reading "Search" whose accessible name is "Search for a value" conforms perfectly well. A button reading "Send" whose name is "submit-form" does not, and saying "click Send" is then unlikely to reach it, which leaves the user hunting for a slower route. The audience is wider than people who cannot use their hands. Pain, fatigue and being able to type only for short stretches all bring people here, and plenty of them use voice alongside a keyboard rather than instead of one.

Voice control, in full →

On this siteRule 2.5.3 Label in Name · Assistive technology, in full

VoiceOver

VoiceOver is Apple's screen reader, built into every Mac, iPhone and iPad, and into the Apple Watch, Apple TV and Vision Pro as well. It speaks the screen and it drives a braille display, which is a first-class output rather than an extra. On a phone it runs on touch gestures, swipe right for the next element and double tap to activate. It also changes what every standard gesture does, which is what makes the first ten minutes of testing feel so strange. Its share tells two different stories. On a desktop it is a distant third, the main screen reader of about one survey respondent in ten, so testing on a Mac is no substitute for testing in JAWS or NVDA on Windows. On mobile it is the opposite. Around seven in ten of the same respondents name an Apple device as their main phone, and VoiceOver is what they use on it.

VoiceOver, in full →

On this siteHow we test with screen readers · How to test with a screen reader

VPAT

Also called: Voluntary Product Accessibility Template, and the filled-in result is an ACR

A VPAT is the blank form vendors fill in to report how their product measures against accessibility standards, one row per requirement, using one of five verdicts. Supports, Partially Supports, Does Not Support, Not Applicable, and Not Evaluated. That last one is allowed only against Level AAA criteria, so a report using it anywhere else is defective by the template's own rules. The completed document has a different name, an Accessibility Conformance Report, and buyers will say VPAT and mean the report. The current template is version 2.5Rev, from April 2025. There is no such thing as VPAT certification, and nobody accredits one. An ACR is a self-declaration, so what makes it worth anything is the testing behind it and the candor of the remarks.

VPAT, in full →

On this siteSection 508, explained · VPAT and ACR, explained · A customer asked for a VPAT

Terms starting with W

Top ↑

W3C

Also called: World Wide Web Consortium

The W3C is the international non-profit that publishes the web's standards, including CSS and, through its Web Accessibility Initiative, WCAG. Tim Berners-Lee founded it in 1994 and more than 330 member organizations work on the standards. It became a legal entity in its own right in January 2023, after 28 years hosted by four institutions, three universities and a European research consortium. The word to watch on anything it publishes is Recommendation. That is the name for a finished W3C standard, and WCAG is one. A W3C Note is a different kind of document on a different track, and it is not a standard. That distinction decides whether the page somebody just sent you is quoting the standard or explaining it. We are independent of the W3C. We explain and audit against their standard.

W3C, in full →

On this siteTheir standard, explained · WCAG versions and levels

WAI

Also called: Web Accessibility Initiative

WAI is the branch of the W3C that develops accessibility standards and guidance. WCAG comes from it, and so does the ARIA specification, which is why you see "WAI-ARIA" written out. Those two come from different working groups inside the same initiative, so they are maintained on separate schedules and can sit slightly out of step on a detail. ARIA is a W3C Recommendation, the same document status as WCAG itself, which does not oblige anybody to use ARIA. What it means is that once you do, its requirements govern how. The techniques, the Understanding pages and the tutorials are WAI's too, and those are informative, meaning they explain how to pass rather than adding anything you have to pass.

WAI, in full →

On this siteWCAG · ARIA

WCAG

Also called: Web Content Accessibility Guidelines, usually said "wuh-cag"

WCAG is the international standard for web accessibility, published by the W3C. It is built in four layers. Four principles at the top, thirteen guidelines under those, then the testable success criteria at three levels of A, AA and AAA, with the technique library underneath as informative guidance. The criteria are the testable part, and a claim rests on them together with the five conformance requirements printed alongside them, at a level that is almost always AA. It is also an ISO standard, which matters more than it sounds, because some national laws cite the ISO number instead of the W3C name. WCAG 2.2 is ISO/IEC 40500:2025 and WCAG 2.0 was 40500:2012, so a procurement form asking for ISO/IEC 40500 is asking for this. In practice, "is this site accessible?" is nearly always being asked as "does it meet WCAG 2.1 or 2.2 Level AA?", and those are not quite the same question.

WCAG, in full →

On this siteAll 55 A/AA rules, explained · Level A, AA, AAA

WCAG versions (2.0, 2.1, 2.2)

WCAG 2.0 arrived in 2008, WCAG 2.1 in 2018 with rules for mobile, low vision and cognition, and WCAG 2.2 in 2023 with nine more criteria and one removal. The counts confuse people because two different bases are in circulation. WCAG 2.2 has 55 criteria at Levels A and AA, and 86 across all three levels, and both those figures are right. W3C states in the standard itself that content conforming to 2.2 also conforms to 2.1 and 2.0, so building to the newest version answers the older ones, with one reporting wrinkle worth knowing about below. The working group's own advice is to build to 2.2 whatever your obligation names. The obligations lag well behind. Section 508 still names WCAG 2.0, the European standard in force aligns to 2.1, and 2.2 is what a site built today should be measured against.

WCAG versions (2.0, 2.1, 2.2), in full →

On this siteThe 2.2 rules, explained · Which law cites which version

WCAG2ICT

WCAG2ICT explains how to apply WCAG to things that are not web pages. Documents that travel off the web, desktop and mobile applications, kiosks, and interfaces built into hardware. It exists because WCAG's own words, web page and user agent, do not carry across cleanly to any of those, so somebody had to work through each success criterion and say how to read it there. Your PDF is still opened by something, and the Note says so in as many words. What changes is how the work divides between the thing you wrote and the thing that opens it. What WCAG2ICT is not is a standard. It is a W3C Group Note, dated 11 December 2025, published on a different track from a Recommendation. Its status line says it is endorsed by the working group that wrote it and not by W3C itself nor its Members. The document says the same thing about itself in plainer words. It provides informative guidance that is not normative and does not set requirements.

WCAG2ICT, in full →

On this siteWhat is digital accessibility · PDF accessibility · Kiosks and closed functionality

Terms starting with Z

Top ↑

Zoom

Also called: Page zoom, text resize

Zoom means enlarging a page, usually with Ctrl or Cmd and the plus key, and two WCAG rules govern what has to survive it. Text has to resize to 200% without losing content or function, and captions and images of text are the two things that rule exempts. Then reflow takes over at a viewport equivalent to 320 CSS pixels wide, or 256 tall for content that scrolls sideways, where reading must never ask you to scroll in two directions at once and nothing may be lost getting there. It does not require the page to become a single column. Those are two different tests, so zooming to 200% can leave a viewport still wider than reflow's target, and two-dimensional scrolling there is not a reflow failure. Browser zoom is the route almost every site takes to the first rule, so an in-page text-size widget is usually work you do not need. It only counts where the text actually reaches 200% and nothing is lost getting there. If your layout clips, overlaps or hides buttons when zoomed, browser zoom has not satisfied anything and that is the failure the first rule was written to catch.

Zoom, in full →

On this siteRule 1.4.4 Resize Text · Rule 1.4.10 Reflow

Missing a term that confused you? Email hello@wcagrules.com. If it tripped you up, it is tripping up someone else, and we will add it.

Find out which of these words describe your site

The free scan opens up to 10 of your pages in a real browser and reports the failures a machine can settle on its own. That is a real answer and a partial one, because a scanner cannot judge whether your alt text says the right thing. The full audit adds an expert review of all 55 Level A and AA rules and a real blind screen-reader user on your key journeys.

Go somewhere useful

Find tools, resources and your workspace.

29 destinations