Why build yet another language learning app?

There are literally hundreds if not thousands of apps. And I’m seeing new ones released pretty much every day.

It isn’t that none of them work. Even Duolingo works to some extent. And many are pretty good. So why create a new one?

Small mistakes compound

I’ve done language learning apps for over 10 years (either as side projects or as the main project). Small mistakes in architecture tend to compound over time.

Although I still think Lingo Champion (my first truly proper app) is still a great app, it’s got some shortcomings when it comes to how everything fits together.

Books are forced into texts and into collections, instead of being kept as books. Which means you’ll have to do a lot of data processing which gets expensive and introduces mistakes.

There is only one meaning (translation) per word. Sure, you can separate them by commas, but that’s semantically not meaningful. This means that the browser extension might split different meanings (based on comma) the wrong way, you can’t practice different meanings separately in flashcards etc.

There are many more such small examples. A lot of people still love the app and I keep developing it further. But at some point (more precisely, beginning of this year), I felt that I wanted to start from a fresh slate and build something new.

Same-same, but different

So, I first played around with the idea of doing a flashcard app for everything. A modern replacement for Anki. I wanted to call it Flippar Marco (for reasons too long to explain here — this name eventually became the name of the Lingo Champion’s chatbot instead). I eventually settled on DuoNinja.

I was pretty pleased with the first prototypes. Approaching language learning from that specific angle revealed a lot of problems I hadn’t thought about before. For example, having different meanings per word matters a lot. Sharing decks matters a lot. Etc.

I then started rethinking everything I had taken for granted for Lingo Champion. For some reason, without much thought, I had forced every type of (text) content into the same reader. It works but it can get awkward. No-one really reads news that way in reality.

There are many little details that need attention from the start. Tokenization. Customization. Pronunciation aides etc. It’s a real pain to add them later.

I also rethought what it means to truly learn a language through audio and videos. So, the video player in DuoNinja is built in a way specific way. The transcript editor (coming soon) will have opinionated design as well.

And I finally figured out a way to do double subtitles in YouTube as well. It’s something I’ve been trying to do for years.

Thinking through the entire app from scratch in the very beginning really paid off, I think. Personally, I’m happy with pretty much every flow — from finding stuff to read and watch to creating Duolingo-style lessons with one click from saved words.

Delete, delete, delete

With AI, it’s very easy to go from zero to some kind of a product. But I think “some kind of a product” is not good enough. It’s just slop. For something so stressful like language learning, it has to be a great product.

So, I took extra care to simplify and remove everything unnecessary. Then come back the next day and remove even more. I played around with prototypes for weeks before I committed.

This is not to say that the app issomehow bare bones. In reality, it beats the best in class in breadth of features in many cases. But the interface does not feel cluttered. The goal wasn’t to keep adding endless features but to work on smoothing the user experience until every transition felt natural.

Languages are still mostly taught wrong

I can’t count how many times I’ve heard something like “You speak seven languages? You must be really good at languages!” I’m very average at best. But I’ve had the good luck of picking them up the right way. Through comprehensible input.

Almost all schools and most apps are stuck in the same old mindset — everything has to be planned for the user and controlled. There is no room for discovery or play (although Duolingo does a good job of pretending it). Much less for real content.

But that’s not how anyone learned their mother tongue. We didn’t do endless grammar lessons. We didn’t drill meaningless dialogues. We didn’t read or listen to stuff we weren’t interested in.

It doesn’t mean that a clear roadmap is meaningless. There’s power in certainty as well. But this can’t turn into tyranny.

So, DuoNinja does have Duolingo-style courses. But you’re not forced to take the lessons in a certain order. And you can also create your own courses.

It does have grammar roadmaps but you can also create your own.

It has recommended YouTube videos and channel but you can also import your own.

It has books and graded readers but you can also create your own.

It has shared Anki-style flashcard decks but you can also create your own.

If you have the same mindset — you like exploring new languages and cultures — then give it a try. It’s starting to have a pretty decent library of content (1000+ video channels, hundreds of books, and tens of thousands of news articles spread across 40+ languages).

Build up your reading and listening fluency first. Know what’s going on in that particular culture. And then jump into more serious grammar study. You’ll then have something to talk about.