Language Access and Accessibility Are One Project, Not Two

In most agencies these are two different programmes. Language access sits with a civil rights officer or a community engagement team and is driven by Title VI. Digital accessibility sits with IT or communications and is driven by the ADA. Different owners, different budgets, different vendors, different meetings.

The documents do not respect that boundary. A public notice has to be readable by a resident who does not read English and by a resident using a screen reader — and, inevitably, by the resident who is both. Run the programmes separately and you will pay twice and still publish files that fail.

Two obligations, one document

The two requirements come from different law and mean genuinely different things.

Title VI of the Civil Rights Act prohibits national-origin discrimination by recipients of federal financial assistance, which has long been understood to require meaningful access for people with limited English proficiency. Title VI itself is untouched — but the executive order that shaped how agencies implemented it for two decades was revoked in 2025, so the framework changed even though the statute did not. What language access requires of public agencies now covers where that leaves you, and how to build a Title VI language access plan covers the plan itself.

The ADA — through the DOJ Title II rule for state and local government, or the HHS Section 504 rule for federally assisted healthcare — sets a technical standard of WCAG 2.1 Level AA for web content and documents. Dates and tiers are in the ADA Title II deadline post.

Neither obligation satisfies the other. Translating a document does nothing for a blind English speaker. Tagging a document does nothing for a Spanish speaker. And here is the part that catches careful agencies: doing both, separately, still fails the person who needs both.

Why sequencing them fails

Say you do the responsible thing in the obvious order. You translate your enrollment packet into Spanish this year, because Title VI is the older obligation. Next year the accessibility deadline approaches, and you send your document estate — now twice the size — for remediation.

Three things go wrong.

You pay for volume you created. Every document exists in two or three languages now, so the remediation bill scales with the number of languages.

The remediation vendor cannot finish the job. They tag your Spanish document competently and cannot write Spanish alt text, cannot judge which passages need language marking, and cannot verify the result with a Spanish screen reader — because they do not read Spanish. You get a structurally tagged file that is still wrong. Alt text in translation covers that specific gap.

Structure gets discarded and re-derived. Translation frequently flattens document structure. The remediation step then rebuilds headings and reading order that existed in your original — work you are paying for twice.

Reverse the order and it is no better: remediate first, then translate, and the translation step strips the tags you just paid for.

What the combined job looks like

The order that actually works is one pass:

1. Fix the source. Real headings, table headers, alt text, and a sane reading order in the original. Everything downstream inherits this, so it is the highest-leverage step and the cheapest.
2. Translate with the structure intact, so headings stay headings and tables stay tables.
3. Tag once, in every language, with document and passage-level language declared.
4. Author alt text in the target language, as content rather than metadata.
5. Verify with a screen reader in each language, by someone who can hear whether it is right.

Note what changes: the expensive, skilled, hard-to-automate steps happen once rather than once per programme. And the verification is done by someone who reads the language, which is the only way step 3 and step 4 can be confirmed at all.

How to combine two programmes that report to different people

The practical obstacles are usually organisational rather than technical.

• Build one document inventory, not two. Whoever owns it, there is one list of documents the agency publishes. That list is the shared artefact.
• Prioritise by consequence, jointly. Emergency notices, forms people must submit, and anything with a deadline come first — for both obligations at once.
• Write one procurement. This is the big one. A solicitation asking for translated and conformant deliverables gets you a vendor who can do both; two solicitations get you two vendors who each do half. How to write accessibility into an RFP includes clause language for multilingual deliverables.
• Fix the publishing process once. The staff who post notices need one checklist covering both, not two.
• Report on both together. A single dashboard showing which documents are accessible in which languages makes the remaining gap visible. Two separate reports make it easy for the intersection to be nobody’s number.

The number worth knowing

The question to ask internally is not “how many documents are accessible” or “how many are translated”. It is:

How many of our documents are usable by someone who is blind and does not read English?

In most agencies that number has never been calculated, and when it is, it is far lower than either individual programme’s figures suggest. It is also the number that a complaint will be about.

Where Taika fits

Taika does both trades, which is why we can do them in one pass. Send a source document and get back tagged, correctly structured files in every language you publish — with alt text in the target language, passage-level language declared, right-to-left reading order handled, and screen-reader verification by someone who reads the language.

Start at accessibility and compliance services to scope the combined obligation, document accessibility for the document estate, or ADA website accessibility for the site itself. For sector-specific starting points: education language access and healthcare language access.

Running two programmes and suspecting they should be one? Request a quote with your document count and the languages you publish in, and we will come back with a single sequence that satisfies both obligations. If you want a data point first, take one of your translated documents and run it through the PDF accessibility checker — it is usually a short conversation after that.