WordPress official languages act compliance: building parallel multilingual translation pipelines
The OL Act does not say “translate the English.” It says equal quality across both languages. That is a CMS standard, and most WordPress bilingual setups silently fail it.
If your French version exists as a Google Translate widget, this post is not the help you need yet. This is for departments with bilingual editorial teams ready to do the work properly.
What equal quality means in a CMS
I have built several bilingual WordPress sites, and the same thing trips every one of them: WordPress is a primary-language-first tool, with the second language bolted on rather than sitting beside the first as an equal. Equal quality means same publish dates, same metadata depth, same image captioning, same internal-link density, same canonical structure across both languages. In practice that is a checklist you can run on any pair of pages. Compare the publish dates; a gap of three weeks tells you the French one lived in a queue. Compare the meta descriptions, because a hand-written English one beside a French one left to the theme's default excerpt is a failure the reader never notices but a reviewer does. Count the internal links, since the English draft collects them over months while the French sits at launch-day count. Every one of those is an editorial-workflow standard, which is why I build it into a WordPress development engagement from the first sprint instead of later, the way the platform nudges you to.
The breaks I find most often are the ones nobody looks at: sidebar widgets that point only at English content, structured-data fragments hardcoded in English, navigation labels that drift apart between the locales. The pages look balanced; the scaffolding underneath is entirely English.
Who attests to parity, and what they look at
The section that usually gets cited, section 41, sits in Part VII of the Act and covers support for English and French minority communities. A bilingual website's obligation sits in Part IV, communications with and services to the public, which requires the two versions to be of equal quality, backed by the Treasury Board policy on official languages. Procurement asks who signed off on that parity, and the answer cannot be “the plugin handles it.” I have watched teams assume WPML or Polylang settled the question, when all those tools actually do is keep the English post and the French post linked; they do not judge whether the two are of equal quality. Someone has to attest to that.
In my experience the attestation review is unglamorous. Someone with both languages picks a sample (the homepage, a service page, the newest three posts, a form) and walks each pair with the checklist above, noting every place the French version is thinner or older. The date gets recorded and the sample is repeated after every content sprint. That is a record a department can put in front of a reviewer, which a plugin settings screen will never be. Ontario's French Language Services Act and New Brunswick's Official Languages Act carry parallel expectations where they apply, which is why the parity question shows up among the questions I ask on a government discovery call.
What a real bilingual editorial cycle looks like
Think of it like a recipe for two cooks working the same dish side by side. The recipe is the editorial brief, read by both before anyone touches a pan: the English writer and the French writer get the same source material, the same image list and the same deadline. Prep happens together, so terminology is settled in a shared glossary before either draft exists, because the fastest way to lose parity is a French writer discovering a new program name in finished English copy. Cooking happens in parallel, with translation memory catching the phrases already decided. Plating is one publish action scheduling both posts for the same minute, metadata and alt text filled in on both. The tasting is the audit walk-through above, done on both pages before either is called finished. The moment the French version becomes a queue downstream of the finished English site, you have already lost the parity the Act asks for.
Working that way costs something, so here is the number. Bilingual editorial training: $6,000 to $12,000 CAD. Ongoing translation memory and review tooling: $8,000 to $15,000 CAD per year. Skipping either: an audit finding at the next review. Over three years the tooling is the bigger line, and it keeps the glossary alive after the people who built it move desks, which they will.