Why a suggested reply is not cheating (it is what keeps you talking)

Watch a beginner in a real conversation and you can see the exact moment it collapses. A question arrives. They understand it. They start assembling an answer, hit a word they do not have, stall — and the silence stretches until someone changes the subject or the session quietly ends.

That failure has almost nothing to do with knowledge and everything to do with time. Given five minutes and a dictionary, they would have produced a fine sentence. They had four seconds.

Conversations die from stalls, not from mistakes

This is worth stating plainly, because most learners have it backwards. A conversation full of grammatical errors runs happily for an hour. A conversation with three long silences ends.

The practical consequence: keeping the exchange moving is more valuable than keeping it correct. You get more practice from twenty flawed sentences than from four perfect ones and three collapses.

What a suggested reply actually does

A one-tap suggestion — Amiko puts it behind the ✨ button — gives you a sentence that would work here, right now, in this conversation, at your level.

Three things happen when you use it.

The conversation survives, which is the immediate point. Instead of a dead exchange you get another turn, and another after that.

You see a model answer at the moment you needed it. This is the part that matters for learning. A phrase encountered in a textbook chapter is filed under "things I read". The same phrase handed to you two seconds after you failed to produce it lands somewhere much stickier, because it arrives attached to your own frustration.

And you learn what a natural answer looks like in that situation, which is often not what you were trying to construct. Beginners frequently attempt a translation of an elaborate sentence from their own language when the natural reply in the target language is four words.

The failure mode: tapping it every time

If you press it on every turn, you are reading, not producing, and reading does not build retrieval. This is a real risk and worth guarding against.

The rule we would give: try first, always. Give yourself five seconds to produce something. Anything — a fragment, a wrong tense, one word. Only then tap.

Those five seconds are the entire exercise. What comes out of them is often bad, and it does not matter; the retrieval attempt is what strengthens the pathway, whether or not it succeeds. Tapping before trying skips the exercise and keeps only the reading.

A better use: try, tap, compare

The strongest version of this habit has three steps.

Produce your own answer first, out loud or in the box. Then look at the suggestion. Then notice one difference — a particle you missed, a shorter phrasing, a verb form you would not have chosen.

One difference per turn is plenty. Over a week of conversations that is thirty specific, personally-earned corrections, each one attached to a moment where you tried and came up short. That is a considerably better teacher than a grammar list.

When to stop using it

You will notice the suggestions starting to match what you were already going to say. That is the signal to stop reaching for it and let yourself struggle a bit longer instead.

The tool is a scaffold for the stage where silence would otherwise end the practice. Once you can carry the conversation without it, it has done its job — and the fact that you no longer need it is the clearest progress measurement the app offers.

You can try this in the browser: start a conversation, set the level to L1, and force yourself to try before you tap.