Neurodiversity

Our AI accessibility engine reviews text for six key areas that impact neurodiverse readers. Each section below explains what we check, why it matters, and how we provide suggestions.


Language of Page

We verify that the overall page language is set correctly. This ensures screen readers pronounce words consistently.

Example Suggestion:

“Detected language is English (US). Please specify HTML lang attribute.”


Language of Parts

Some content has mixed languages (e.g., an English page with a French quote). We detect these language spans and recommend marking them.

Example Suggestion:

“Bonjour le monde” should be tagged as French for proper pronunciation.


Unusual Words

Complex or uncommon words can be flagged for clarification or replacement.

Example Suggestion:

Flag: pulchritude
Suggestion: replace with beauty or provide a tooltip definition.


Abbreviations

Abbreviations are often unclear without context. We suggest expanding them on first use.

Example Suggestion:

The word “FBI” is flagged.
Suggestion: use Federal Bureau of Investigation (FBI).

Even familiar acronyms can be confusing for some readers. Expansion improves clarity.


Reading Level

We analyze text readability using the Flesch-Kincaid formula. Content that is too complex may exclude neurodiverse readers.

Example Suggestion:

Before (Grade 14):

“In consideration of the multifaceted circumstances surrounding this issue, it is imperative that stakeholders collaborate expeditiously to formulate a comprehensive resolution.”

After (Grade 8):

“Because this issue is complex, it’s important that everyone works together quickly to find a clear solution.”

This makes the text easier to process while keeping the meaning intact.


Pronunciation

Screen readers can mispronounce certain words or names. We flag those cases and suggest phonetic guidance.

Example Suggestion:

Word: “GIF”
Suggestion: mark pronunciation as “jiff” or “gif” (hard G) depending on intent.


To reduce noise, our engine deduplicates findings on every page. Each issue is shown only once, even if it appears multiple times. This keeps feedback clear and avoids overwhelming the writer.