# Native HTML Semantics Beat Re-Implemented ARIA ## Claim Reach for a native HTML element whose semantics and behavior already match the need, because re-purposing a generic element with ARIA roles tends to add errors rather than accessibility. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds for interactive controls with native equivalents — buttons, disclosures, form fields, links — where the browser already provides focusability, keyboard activation, and the correct role. - Breaks where no native element provides the needed pattern; targeted ARIA (live regions, `aria-expanded` on an already-native control) remains legitimate there. - Applies with extra force to LLM-generated markup, since the failure mode (scraping the ARIA Authoring Practices Guide for "patterns") systematically reaches for the ARIA-heavy version even when a native element would work. ## Evidence The W3C's own first rule of ARIA states it directly: use a native HTML element with the needed semantics before re-purposing a generic element with a role. The canonical anti-pattern is `<div role="button">`, which requires manually adding `tabindex`, keyboard handlers, and the role — everything a real `<button>` provides for free. This isn't just a style preference: pages with higher ARIA-attribute density show measurably more detected accessibility errors, consistent with the WebAIM Million's recurring finding. ## Links - [[3. Resources/Literature Notes/ARIA, anti-patterns, and you]] — because evidence-for; source Literature note holds the full claim-atom breakdown. - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — because applies-in; LLM-generated markup is a scaled instance of this exact failure mode. - [[3. Resources/MOCs/Accessibility & Disability Inclusion MOC]] — a build-time decision that prevents remediation rather than performing it - [[3. Resources/Permanent Notes/A Shared Accessible Component Library Makes the Accessible Choice the Default]] — has evidence from: the accessible choice must be what a developer gets by typing the obvious thing (giffgaff, LeadDev). ## Tests / falsifiers - Crawl a sample of theme/block output, correlate ARIA-attribute density against automated-checker error counts, controlling for page size — if density doesn't track with errors in this sample, the empirical backbone weakens. - A/B a component-building task: one cohort builds by pointing an LLM at the APG, another builds native-first; count WCAG failures and manual-test defects on matched components. ## Maintenance - Last tended: 2026-08-16 - Next review: 2026-11-16 ## Source - [[3. Resources/Readwise/Articles/ARIA, anti-patterns, and you]] — David Bushell's correction of the APG-as-cookbook misreading and the case for native-first HTML.