Since 2009

We are the people who translate the software you already use.

OpenTranslators started as a mailing list of fourteen volunteers arguing about a Joomla glossary. It is now a working collective of translators, reviewers and engineers who localize open source projects for a living — and for free, depending on the project.

Diagram: translators offer a language, developers submit software, an open platform connects them.
The original diagram from our first site. Fifteen years on, the model has not changed.
Principles

Six things we will not compromise on

These are not values on a poster. Each one is a rule we have actually turned work down over.

A named reviewer, always

Every string has a translator and a second person who signs it off. No self-review, no exceptions, even under deadline pressure.

Context before words

A translator working from a bare string list is guessing. We supply screenshots, comments, character limits and a glossary first.

Credit is not optional

Contributors are named in the release, in the repository and on this site. Unpaid does not mean anonymous.

Fix the source, not the symptom

If a string is untranslatable because of how it is built, we send a patch to the code. We do not paper over it in forty locales.

Machines assist, people decide

We use machine translation where it earns its place, and we say so openly. A human always makes the final call.

Everything we learn is public

Glossaries, style guides, tooling and post-mortems are published under open licences. Including the mistakes.

History

How we got here

A mailing list and a glossary argument

Fourteen volunteers set out to standardise Joomla extension terminology across five languages. The glossary outlived the argument.

First shared platform

Teams moved off spreadsheets and email attachments onto a shared translation platform. Throughput roughly tripled in a quarter.

The reviewer ladder

We formalised how a new translator becomes a reviewer, and then a team lead. Contributor retention stopped being luck.

Continuous localization

String extraction moved into CI. Translators started seeing new strings the day they were merged rather than the week before release.

Quality gets a number

We adopted an MQM-based scorecard so quality conversations could be about evidence instead of taste.

Six hundred and counting

Forty-two locales, one hundred and eighteen projects, and a publishing habit we have no intention of breaking.

Questions

Things people ask us

Both, honestly. Paid client work funds a small core team; that core team supports volunteer language teams working on projects that could never pay. If your project is community-run and unfunded, say so when you write to us — we reserve capacity for exactly that.

Forty-two locales have an active team with a named lead and at least one reviewer. Another twenty or so have a single translator and no review capacity — we will tell you plainly which category your target locale falls into rather than quietly shipping unreviewed work.

For high-volume, low-risk content with human post-editing, yes, and we label it. For interface strings, legal text, error messages and anything a user acts on, no. We wrote up the full reasoning in our article on machine translation and post-editing.

Engineering work is quoted per scope after an audit; translation is per word with a separate review rate. Community open source projects are quoted at cost or free depending on our capacity that quarter. We publish the rate card during scoping, not after.

You do, under whatever licence your project uses. Translation memories and glossaries we build with you are yours to export at any time, in a standard format, without asking us first.
The core team

Who you will actually talk to

Small on purpose. The people who scope your project are the people who do the work.

Matt Sorensen

Matt Sorensen

i18n engineering lead. Fifteen years of pulling hard-coded strings out of other people's codebases. Writes the audits and most of the lint rules.

Tess Okonkwo

Tess Okonkwo

Terminology & QA lead. Owns the glossaries and the scorecard. If your translation got a number, Tess decided how it was calculated.

Hilde Sandvik

Hilde Sandvik

Community programme manager. Recruits, onboards and mentors language teams. Keeps the reviewer ladder honest.

GnuMax

GnuMax

Platform & tooling. Keeps the extraction pipelines, CI hooks and the public glossary API running.

Ready to speak to the whole world?

Tell us about your project and we will come back within two working days with a scope, a timeline and the names of the people who would do the work.

Start a project See our services