Building a mission-driven app for language preservation and revitalization

"By the middle of the next century, we will be losing our linguistic heritage at the rate of 26 languages each year—one every two weeks. If we do not tackle the problem of language loss, more than half of all languages will become extinct in the next 100 years." - The Language Conservancy

Languages are disappearing at an alarming rate

Most of the world's languages are at risk of disappearing within the next few generations. To prevent catastrophic language loss, we urgently need to prioritize language preservation, revitalization, and the creation of learning resources for endangered and less-commonly-taught languages.

This is why I'm creating Elm, an app that prioritizes building learning experiences for less-commonly-taught languages.

With Elm, my mission is to build an ecosystem to empower the world's under-served language communities. This ecosystem will start its life as a web app, and eventually expand into a mobile app.

What makes Elm unique?

1. Elm prioritizes creating learning resources and experiences for endangered languages, classical languages, and less-commonly-taught languages.

Many learning programs, apps, and platforms are doing a fantastic service to learners interested English, Spanish, Chinese, and other frequently-taught languages. But I see few if any resources for learners interested in languages like Armenian, Nahuatl, Amharic, and Mongolian. 

With this project, I'd like to reshape the language learning landscape by creating opportunities for people interested in learning these languages, whether they are heritage speakers, or interested in learning these languages for personal or professional reasons.

2. Elm's UX will emphasize tracking language skills and proficiencies, enabling fine-grained personalization.

Readers who are in the language education space will likely be familiar with the CEFR. It's a 6-point scale for describing a learner's language ability, ranging from A1 (beginner) to C2 (near-native), and is commonly used as an international benchmark.

While CEFR levels can provide a holistic sense of a learner's proficiency, it's difficult to infer exactly what kind of knowledge a learner has based just on their level, or from its CEFR description. For example, B1 includes this description:

• "[The learner] can deal with most situations likely to arise while traveling in an area where the language is spoken."

From a description like this, I can get a general sense of a learner's proficiency, but I can't distinguish that learner from other B1 level learners. And crucially, I can't personalize a learning experience based on such a broad generalization.

To enable greater learning personalization, I'm taking a much more analytical/reductionist approach by tracking individual learning objectives. A learner's knowledge of each glyph, sound, word, grammatical construction, etc. will be tracked, allowing for a much finer-grained view into a learner knowledge. These learning objectives can then be bundled up to CEFR levels, allowing for a more familiar snapshot of a learner's progress.

One major upshot of this approach to granularly describing a learner's language knowledge is that it enables a much greater degree of personalization.

More about this can be found in my portfolio case study on this project.

3. Elm's learner experience (LX) will feature dynamic learning scenarios

In my experience, most education apps — not only language learning apps, but education apps generally — are built around lessons as the core experience. 

Lessons tend to consist of static, pre-defined content. And when you repeat a lesson, you get the same content, usually delivered in the same order. This kind of experience is consistent and predictable, which can be helpful for learners. But it is also rigid and difficult to personalize. With this kind of structure, personalization is often achieved by skipping lessons by testing out of them via placement tests.

To rethink this status quo, I'm creating what I'll refer to as "scenarios". Much like lessons, scenarios are bundles of learning content. However, unlike lessons, scenarios will be dynamic. They will adjust to the learner's background knowledge (demonstrated and inferred), and thus be personalized in that sense.

This of course doesn't mean that scenarios will consist of randomized content. Instead, scenarios will have specific learning objectives that can be taught in open-ended ways. 

To give a simple example: Imagine two learners are in a scenario whose objective is to teach the names of common fruits. One learner has already learned the past tense, while the second learner has not. For the learner who knows the past tense, such constructions may occur in this scenario, whereas they would not for the second learner.

Much more can be said about scenarios and personalization, but I'll conclude this section with a screenshot of my most recent prototype, demoing three tabs of a learning scenario: (a) the learning activity; (b) the scenario progress overview; and (c) learning resources for the current activity.

And as an aside, below is a simple example of two different learning activities. Both are multiple choice probes, but each tests a slightly different bit of knowledge: (a) mapping from a glyph in the target language to its transliteration in English; and (b) mapping from a sound in the target language to its written representation in the target language.

When will Elm launch?

Finally, the pressing question: when will this app see the light of day?

My plan is to launch a private beta later this year (Q4 2023). 

Operationally, this means that I'll launch a very demo-ish version of the app, invite a small number of people to test it, and work to incorporate all the feedback I can before launching a public beta in early 2024.

If you'd like to be a part of the private beta, please reach out and let me know. I'd love to have your input!

Thanks for reading!