Everyone selling you a language app is selling you the wrong half of it.

We spent a week reading everything we could find on whether gamification actually makes people learn a language faster, fully expecting to come out the other side calling it decoration. It is not decoration. The research is clear, and we were wrong about it. Gamified learning produces a real gain in what people actually learn, and two separate reviews, working from completely different piles of studies, landed on roughly the same size.

What we did not expect was where the gain comes from.

Take the points away. Take the badges, the streak, the little celebration sound. What is left is a student sitting down on a Wednesday night and pulling English out of their own head instead of reading it off a page. That one thing is worth more than everything else in this article combined, and it works whether or not there is a leaderboard anywhere near it.

The game is the door. The practice is the room. Almost everybody in this industry sells the door.

So what is the game actually giving your students?

Belonging. Not capability. Once you see it that way, a lot of confusing things about gamification stop being confusing.

Think about what actually runs through a student's head when they open an app with a league in it. They are not thinking about the past perfect. They are thinking about the person two places above them, and whether they can catch her by Sunday. The game puts your student in a room with other people. That is a real and valuable thing, especially for someone learning alone at night. It just is not the thing you thought you were buying.

That is not only our impression. A 2024 review by Lingyun Li and colleagues measured this need by need and found the same split: a large jump in how connected learners felt, and almost nothing on the sense of actually getting better.

Duolingo says it too, from the other direction, and we think this is the most honest thing in the entire field. Every number they publish about streaks is a number about people coming back. Retention up 14 percent. More likely to return after the weekend. Not one of them is about knowing more Spanish. They are not hiding anything. It is just that attendance is easy to measure and learning is not.

So when someone tells you their app is proven to work, ask what it was proven to do. Usually the answer is: proven to make people open it.

Then what actually moves the English?

One thing above all others, and it is boring to look at.

Make them produce the language from memory, without the page in front of them. Not recognize it. Not reread it. Produce it.

The clearest demonstration we know of comes from Jeffrey Karpicke and Henry Roediger, who gave learners forty Swahili words. One group kept getting tested on them. Another kept rereading them. A week later, the tested group remembered about 80 percent. The rereading group remembered about 36.

Same words. Same total time. More than double.

The detail we cannot stop thinking about is what the learners predicted. Every group guessed they would remember about half. Nobody could feel which method was working. The one that felt harder and less satisfying was the one that worked, and the one that felt smooth and productive was the one that failed a week later.

That should make you suspicious of your own instincts, and of ours. It should also make you suspicious of any activity you judge by how well it went in the lesson. Rereading beats testing after five minutes and loses badly after a week, so anything measured on the same day will quietly select the worse method every time.

Then there is the calendar, which is free. Spreading the same practice across days instead of cramming it lifts what people remember by around ten points. Same material, same effort, different days.

That is the acceleration. Not the badge.

So are the points pointless?

No, and this is where people overcorrect.

Retrieval and spacing only work if the student actually sits down on Wednesday. And nobody sits down on Wednesday because a study told them to. They sit down because something small and human is pulling them back. That pull is the job the game does, and nothing else does it as cheaply.

The trouble starts when the game stops being the door and becomes the destination.

You can watch it happen in Duolingo's own community, where researchers read through thirty thousand comments and found people replaying the very first lesson over and over just to farm points. One learner said out loud that they did not even care what they were learning. That is a student with a beautiful streak and an empty head, and every gamified system produces some of them.

It also fades. A full semester of a gamified course, compared against the same course without the leaderboards and badges, ended with the gamified students less motivated, less satisfied, and scoring lower. Sixteen weeks is long enough for the shine to come off. A design that delights in September can grate by November.

Our read: use the game to open the door, and put your real effort into what happens after they walk in.

What about a leaderboard when you have a hundred students?

Honest answer first, because nobody else will give it to you: this has not been studied. A researcher proposed exactly this experiment back in 2019, comparing leaderboards at ten, twenty and fifty people, and as far as we can find the result was never published. Anyone telling you the ideal leaderboard size has been measured is guessing.

What we would do, and why.

Keep the groups small, because more competitors make people try less, not more. That sounds backwards until you picture it. Beating ten people you can name feels possible. Beating a thousand strangers is not a competition, it is weather.

And rank against a target, not against each other. When students are ranked against classmates, the ones at the bottom get measurably worse. When the goal is a fixed number, position stops predicting performance. In that same work, the students at the bottom asked for the ranking to be anonymous.

They told the researchers what they will never tell you.

Try this in your next class

  • End the lesson by asking, not reviewing. Close the notes, make them produce the word out loud. Five minutes of that is the strongest thing in this article.
  • Never judge an activity by how it went today. Check what survived a week later. That is the only number that means anything.
  • Say the progress out loud, specifically. Not "great job." Something closer to "your past tense is steadier than it was a month ago."
  • Split a big group into small tiers, and give them a target instead of a rival.
  • Watch who goes quiet after the leaderboard appears. Those are the students you are losing, and they will not raise their hand about it.

The honest bottom line

Gamification works, and we would rather say that plainly than be fashionable about it. Anyone calling it pure decoration has not read the research.

But it works mostly by getting your students back in the room and making them feel like they belong there. The actual learning comes from what happens once they arrive: pulling the language out of their own memory rather than reading it again, in short sessions, spread across days.

If you have limited time and energy, and you do, spend it there. The points are the door handle. They are not the room.

That split is also how we built our own product, for what it is worth. If you run a school and want the practice to keep happening between your classes without you personally chasing a hundred and fifty people, that is the problem Ocean English exists to solve.

Where this comes from