Most of our catalogue is what you would expect: Spanish, French, Japanese, Korean, Arabic — the languages with millions of learners and no shortage of tools. Alongside them sit a handful that raise eyebrows. Cheyenne. Ojibwe. Mizo. Hakha Chin. These get a fraction of the downloads of the Spanish dictionary, and they take disproportionately more work to build.
They are also, by some distance, the ones we get the most email about.
What "under-resourced" actually means
It is easy to assume it means "few speakers". It does not. A language is under-resourced when little about it exists in machine-readable form — few digitised texts, limited dictionary work, no substantial corpora for software to learn from.
Those two things come apart more often than people expect. Mizo, spoken in the Indian state of Mizoram, has several hundred thousand speakers, official status at state level, and everyday use in education and broadcasting. It is a healthy language by any measure. It is still thinly represented in digital tools compared with languages of similar size in Europe, because resourcing follows commercial attention rather than speaker counts.
The four in our catalogue
| Language | Family | Mainly spoken in | Situation |
|---|---|---|---|
| Cheyenne | Algonquian | Montana and Oklahoma, USA | Severely endangered; most fluent speakers are elders |
| Ojibwe | Algonquian | Great Lakes region, USA and Canada | Endangered, with active immersion schooling |
| Mizo | Tibeto-Burman (Kuki-Chin) | Mizoram, India | Healthy, official at state level, thin digital coverage |
| Hakha Chin | Tibeto-Burman (Kuki-Chin) | Chin State, Myanmar; large diaspora | Substantial community, including refugee populations abroad |
Four languages, four different situations. Lumping them together as "rare languages" obscures the fact that a Cheyenne dictionary and a Mizo dictionary serve almost entirely different needs.
Who actually opens these apps
The usage looks nothing like the Spanish dictionary, where a typical user is an adult learner working through vocabulary. Here the pattern is:
- Community members and heritage speakers. People who grew up hearing the language from grandparents, understand more than they can produce, and are reaching back toward it.
- Language programmes and immersion schools. Ojibwe in particular has serious immersion education, and teachers and students need quick reference during the day.
- Diaspora families. Hakha Chin has significant communities resettled far from Chin State, where parents are trying to keep a language alive with children growing up surrounded by English.
- Documentation and academic work. A smaller group, but the one most likely to send us detailed corrections.
Why they are harder to build
For a major language, dictionary data is a solved problem — extensive published lexicography exists and the work is engineering. For these languages almost every step is harder:
- Sources are scarce and sometimes disagree. Where two dictionary traditions exist, they may use different orthographies, and choosing between them is a decision with consequences for the community.
- Orthography may still be settling. Writing systems for languages primarily transmitted orally are often recent, and conventions vary by region, generation and denomination.
- Structure does not map onto English. Algonquian languages such as Cheyenne and Ojibwe are polysynthetic — a single word can carry what English needs a clause for, with meaning built from stacked pieces. The idea that a dictionary is a list of words paired with translations fits these languages poorly, which is a linguistic problem before it is a software one.
- There is no automated fallback. With a widely-supported language you can lean on existing machine translation to sanity-check. Here there is nothing to check against, so errors persist until a speaker notices.
This is why we ask for corrections. For the smaller languages in the catalogue, email from speakers is often the best quality signal available. If you use one of these dictionaries and find something wrong, telling us genuinely improves it — there is no larger dataset we could consult instead.
Why offline is not a marketing feature here
For a Spanish learner in a city, offline access is a convenience. For these languages it is frequently the difference between a usable tool and an unusable one. Rural reservations in Montana and the Great Lakes, hill districts in Mizoram and Chin State, and households where mobile data is a considered monthly expense all share the same property: connectivity cannot be assumed.
A dictionary that needs a connection is missing exactly where the language is spoken most.
They will never pay for themselves
The Cheyenne dictionary will not recover its development cost. Neither will Ojibwe. That is fine — the rest of the catalogue covers it, and a studio that publishes sixty apps can afford a few that exist for other reasons.
There is also a straightforward argument for it. Every language carries knowledge that exists nowhere else: ways of describing landscape, kinship, weather and time that do not survive translation. When a language stops being spoken, that goes with it. A dictionary app is a small contribution against that, but it is not nothing, and it costs us far less than it is worth to the people using it.
Browse the full dictionary catalogue
45 offline dictionary pairs, from Spanish and Japanese through to Cheyenne, Ojibwe, Mizo and Hakha Chin. All free, all working without a connection.
Open the dictionary appFrequently asked questions
What makes a language under-resourced?
It is not about speaker numbers alone. An under-resourced language is one with little machine-readable material — few digitised texts, limited dictionaries, no large corpora. A language can have hundreds of thousands of speakers and still be under-resourced if almost nothing exists about it in digital form, which is the situation for a great many languages worldwide.
Who uses a Cheyenne or Ojibwe dictionary?
Mainly community members rather than foreign-language hobbyists. Language programmes, immersion schools, families reconnecting with a heritage language, and students in tribal college courses. The usage pattern differs sharply from a Spanish dictionary — sessions are often shorter and more focused on specific words that came up in a class or a conversation with an elder.
Why is offline access particularly important for these languages?
Because the communities where they are spoken frequently have patchy or expensive connectivity — rural reservations, remote hill districts, and areas where mobile data is a real household expense. A dictionary that requires a connection is unavailable exactly where and when it is most needed.