Some words look the same when written but mean different things depending on how they are pronounced. When pronunciation determines meaning, users who rely on assistive technologies, or who have difficulty converting written text into spoken or meaningful language, may not be able to understand the content without additional help.
Screen readers usually rely on built-in pronunciation rules and dictionaries. They do not interpret meaning from context the way humans do. When a word is ambiguous, assistive technologies may pronounce it using a default reading that changes the intended meaning.
When pronunciation is essential to understanding, authors must provide a way for users to identify the correct pronunciation.
Note: This requirement applies only to Level AAA conformance. It is not required for Level A or AA.WCAG Success Criterion
3.1.6 Pronunciation (Level AAA)
A mechanism is available for identifying the specific pronunciation of words where the meaning of the words, in context, is ambiguous without knowing the pronunciation.
This is a Level AAA success criterion. While optional, it is especially relevant for educational, linguistic, academic, and multilingual content, where pronunciation directly affects meaning.
Key Terms (Quick Definitions)
Heteronyms
Words that are spelt the same but pronounced differently, with different meanings
(e.g., record as a noun vs record as a verb).
Mechanism (WCAG meaning)
Any consistent, reliable way for users to access information, such as inline text, annotations, audio, or links, provided it works with keyboards and assistive technologies.
Intent of the Success Criterion
This criterion intends to ensure users can understand content when the meaning depends on pronunciation.
WCAG does not require pronunciation guidance for all words. It applies only when:
- A word or character has multiple pronunciations
- Each pronunciation results in a different meaning
- The correct meaning cannot be reliably determined from context alone
In these cases, relying on default screen reader pronunciation can mislead users rather than help them.
WCAG does not require authors to force screen readers to pronounce words a certain way. It requires providing users with a reliable way to identify the intended pronunciation.
When Pronunciation Becomes a Problem
Pronunciation issues commonly arise with:
- Heteronyms
- desert (abandon) vs desert (arid region)
- Proper names with non-obvious pronunciation
- Characters with multiple readings (e.g. kanji in Japanese)
- Shared scripts across languages with different pronunciations
- Academic or linguistic content where pronunciation signals meaning
If pronunciation affects meaning, and context alone does not resolve the ambiguity, support is required.
What WCAG Means by a “Mechanism”
A pronunciation mechanism may include:
- Inline pronunciation guidance (textual or phonetic)
- Ruby annotations (commonly used in Japanese)
- Links to audio pronunciation files
- Glossary entries that include pronunciation
- Parenthetical pronunciation notes
The mechanism must:
- Be available where the ambiguous word appears
- Be reachable by keyboard
- Be exposed to assistive technologies
- Provide the intended pronunciation in that context
WCAG does not mandate a specific technology or markup.
Example
English Heteronym – Textual Mechanism
Please record (/rɪˈkɔːrd/) the meeting so we can keep a record (/ˈrɛkərd/).
This provides explicit pronunciation where meaning depends on it.
Japanese Ruby Annotation
<ruby>生<rt>なま</rt></ruby>Ruby annotations are widely supported by assistive technologies and are the preferred mechanism for showing readings in Japanese.
Note: In Japanese contexts, ruby is typically described as providing the reading rather than pronunciation.
Common Implementation Approaches (With Tradeoffs)
Ruby (<ruby>)
Best for CJK languages; widely supported by screen readers.
Parenthetical pronunciation
Simple and reliable, but can clutter dense text.
IPA (International Phonetic Alphabet)
Highly precise, but inaccessible to many users without training.
Audio pronunciation links
Useful, but must be keyboard-accessible, clearly labelled, and not auto-play.
No single approach is universally best. The correct choice depends on language, audience, and context.
What This Criterion Requires
If a word’s meaning depends on how it is pronounced, and users cannot reliably infer that pronunciation, you must provide a way for users to learn how it should be pronounced.
Users should not be forced to guess, infer from unclear context, or rely on default screen reader output.
If pronunciation affects meaning and users have no way to access it, the criterion fails.
What Does Not Count as Meeting This Criterion
- Assuming context is always sufficient
- Relying on the default screen reader pronunciation
- Providing pronunciation only visually
- Hiding pronunciation in unlinked glossaries
- Using IPA without explanation
- Assuming familiar spelling implies clear meaning
Common Author Mistakes
- Assuming pronunciation “doesn’t matter online”
- Expecting assistive technologies to infer meaning
- Providing pronunciation only once, far from usage
- Using hover-only solutions
- Treating IPA as universally understandable
Relationship to Other Criteria
- 3.1.3 Unusual Words — defines jargon and idioms
- 3.1.4 Abbreviations — expands shortened forms
- 3.1.1 Language of Page — identifies the language
- 3.1.2 Language of Parts — marks language changes
- 3.1.5 Reading Level — addresses overall text complexity
Pronunciation focuses specifically on spoken ambiguity, not vocabulary difficulty or language identification.
WCAG Techniques
WCAG discusses this requirement in techniques such as:
- G121 — Providing pronunciation guidance for ambiguous words
- G62 — Using ruby annotations to provide pronunciation
These techniques are examples, not requirements.
How to Test
- Identify words whose meaning changes based on pronunciation
- Ask: Would incorrect pronunciation change the meaning?
- Check that a pronunciation mechanism exists at the point of use
- Test with screen readers (e.g., NVDA, VoiceOver)
- Verify keyboard access and discoverability
If users hear the wrong meaning and have no way to correct it, the criterion is not met.
Benefits
This success criterion benefits people who:
- Are blind or have low vision
- Use screen readers or text-to-speech
- Have reading or language-processing disabilities
- Are learning the language or reading unfamiliar scripts
Providing pronunciation guidance removes ambiguity and prevents misinterpretation.
Wrapping Up
WCAG 3.1.6 applies only when pronunciation determines meaning, and users cannot reliably infer it. When pronunciation is essential, providing a clear mechanism ensures content is understood as intended. While this is a Level AAA requirement, applying it prevents misinterpretation and improves claritye pecially in multilingual, educational, and linguistic content.
