Endangered Languages Project
A platform for communities to document endangered languages and for researchers to explore and verify language data.
- UX research, Product strategy, UI design, WordPress engineering
- ~35 proposals
- 11
- 8 + 6 months
The brief: one platform, two very different audiences
The Endangered Languages Project (ELP) needed one platform to serve two audiences with almost opposite needs.
For language communities, it had to feel personal and welcoming: a place to share recordings, writing systems, oral histories and resources, and to connect with other people doing the same work. Contributing should not feel like filling in a database form. Community members are adding to something they care deeply about.
For researchers, the same platform had to feel structured and trustworthy. They needed to browse verified data across hundreds of languages, understand how endangered each one is, compare languages, and know where every figure came from.
The challenge was designing one experience that supported both cultural ownership and academic rigour without feeling like two products stitched together.
How we won the work
ELP received nearly three dozen proposals for its new website and Revitalization Helpdesk. They chose RogueUX for our “demonstrated experience in building engaging user communities, supporting multilingual websites, serving international audiences, rigorous UX testing and design, and doing it all at a competitive price point,” and selected our higher-end budget option, at $108,000 USD.
We staffed the project the same week: a product lead who also carried senior product design, a senior software engineer and a junior product designer, working alongside ELP’s communications team.
Product lead and senior product designer
Ran the engagement, led research and design.
Senior software engineer
Database, back end and the WordPress build.
Junior product designer
User stories, wireframes and usability testing.
Communications Director
Product owner on the client side.
Communications Coordinator
Content, community review and admin training.
We redesigned and rebuilt the platform on WordPress over eight months, then came back for a six-month second version.
The core problem: one number, two very different readings
Both audiences met at the same artefact: the vitality score shown on every language page.
A vitality score rates how at risk a language is of disappearing, based on factors like the number of speakers and whether the language is still passed down between generations. ELP’s admin team and community members both entered this data through WordPress, and the score had to surface clearly on each language’s public page.
That created two design problems at once:
- For researchers, the score needed to be scannable and comparable. Someone browsing hundreds of entries had to read relative urgency at a glance, and trust where the number came from.
- For community members, the same number could land very differently. For someone who speaks the language, a low vitality score is a statement about the survival of something deeply personal. The interface could not treat it like a stat on a dashboard. It had to be respectful rather than clinical, while staying accurate.
It was the problem we kept coming back to, through the first version and into the second.
Research: ownership and provenance
We kicked off with lightning talks and How Might We sessions with the ELP team, then sent a Typeform survey to users and stakeholders. After reviewing the results we ran five stakeholder interviews and six user interviews over Zoom, to hear both audiences directly.
The clearest tension to come out of the research was around ownership and trust. Community members wanted their language’s page to feel like a representation of their language and culture, closer to a personal or cultural profile than a record in a database. Some did not trust outside researchers documenting their language.
Researchers wanted almost the opposite: provenance, structure, comparability and confidence in the data.
How can one interface feel personal enough for communities to take ownership of their language, while remaining rigorous enough for researchers to trust the information?
Both needs had to be served by the same interface patterns. That shaped the platform’s visual language: informative and structured, but approachable enough to feel human.
From research to epics
Discovery
Lightning talks, a survey, 11 interviews
Workshops
Personas, journeys, user stories, card sorting
Design
Wireframes, high-fidelity screens, a live prototype, usability tests
Build
Database, back end, front end in WordPress
Launch
Admin training and handover documentation
ELP and its users reviewed the work at every step.
From the research we developed three personas, to keep the different needs concrete throughout the design.
Cultural Carlos
A member of an endangered language community. Motivated by pride and preservation; needs low-friction ways to contribute knowledge, media and resources.
Research Rebecca
A language researcher. Motivated by data integrity; needs to browse, compare and trust where information came from.
Helpful Helena
A language preservation volunteer. Motivated by contribution; sits between the other two.
We mapped each persona’s journey with user flow diagrams, storyboards and journey maps, which showed us where the audiences’ goals overlapped and where they diverged, and so which parts of the product could share patterns and which had to behave differently.
User stories then translated those needs into candidate functionality. ELP stakeholders, language researchers and our designers all wrote as many stories as they could think of; we cleaned them up and grouped related stories into epics such as Language Page, Create a Language, Contribute Resources and Explore Language Data. A card-sorting workshop shaped the site map.
The first version: it felt too “form-like”
Each epic was translated into a set of wireframes and shared with the team and with users for feedback.
Our early wireframes tested a single, dense layout designed to make a large amount of language data available in one place. Community reviewers told us it felt too “form-like,” which was exactly what we were trying to avoid.
That changed the approach. Rather than treating the language page as a database record, we began treating it as a living profile of the language and the community behind it, with clearer hierarchy and more distinct sections. We moved from hand-drawn wireframes to high-fidelity responsive screens and a live prototype in Webflow, and put the prototype through usability testing before building it in WordPress.
The first version was a full redesign and rebuild, from restructuring the resources database to integrating forums and translation management.
Language map
Browse languages by where they are spoken.
Language pages
A page for each language, with its resources and data.
Resource library
A restructured resources database with search and filtering.
Community
A member directory and Discourse-powered forums.
Translation
Translation management for publishing in other languages.
Training and documentation
So ELP’s team could run the platform themselves.
We built the platform in WordPress using Gutenberg blocks, which gave the ELP team control over the content and structure of the site. They update and manage the experience visually, without relying on developers for everyday changes.
The second version: from database entry to online home
The language page in the first version still read like a database entry: data first, with the endangerment level near the top. For the second version we redesigned it to be light and playful, so it felt less like a record in a catalogue and more like an online home for the community.
- The endangerment level moved down the page. It is still there, but it is no longer the first thing a speaker sees about their own language.
- Technical data moved to its own tab. Classification, language code, orthography and data downloads moved to an Info tab, so the overview leads with people, resources and events.
- Empty states became invitations. “No users” became “No ELP members have joined this language page yet. Sign up?” and “Members” became “Contributors.”
- More reasons to come back. Communities can create and share events for their language, organisers earn a “Host” badge on their profile, and people can personalise their profile colours.
Rigour, one tab away
Moving the data off the overview did not mean hiding it. The second version made it more rigorous.
A calculated score. Rather than showing raw data, vitality became a calculated score in which intergenerational transmission counts double.
Intergenerational transmission
Absolute number of speakers
Speaker number trends
Domains of use
The total is divided by the highest possible score for the factors that have data, then scaled to 100, so a language with gaps in its data is not penalised for them.
Seven levels on the map. The language map groups vitality into seven levels (at risk, vulnerable, threatened, severely endangered, dormant, awakening and unknown), each with its own pin. Vitality can be toggled on or off, and a second toggle shows revitalisation programmes, so the map shows where communities are bringing languages back, not only where languages are declining.
Every data point cites its source. The Info tab labels each source “Information from: [citation],” linked whenever a URL exists, and makes room for longer bodies of text.
Contributions are reviewed. Community-submitted events and revitalisation programmes are saved as drafts for ELP’s admins to review and publish, so contributing stays easy without compromising the data.
The result was one page that served both audiences: an online home up front, with rigour one tab away.
Before and after
Drag across each screenshot to compare the original site with the platform we designed and built.
The result
A single platform that supports two fundamentally different ways of engaging with endangered languages. Community members document and share the resources, stories and knowledge connected to their language; researchers explore structured information across hundreds of languages with clearer sourcing and context. One interface serves both: structured enough to support research, and personal enough to give communities a sense of ownership.
See it live: endangeredlanguages.com