What if we just started… talking?
On knowing the words, finding them too late, and making room for speaking.
Picture this. You’re buying a coffee in German.
You’ve ordered. They’ve understood. Things are going suspiciously well.
Then comes a follow-up question. You understand that too. You even know what you want to say.
But you can’t quite get it out.
The person behind the counter switches to English. You finish the transaction. Somewhere outside, your answer finally arrives.
It’s tempting to take that moment as evidence that you haven’t learnt much. But there’s another question worth asking: what, exactly, have you been practising?
When you listen or read, you’re comprehending language someone else has supplied. Speech production asks you to retrieve words from memory, organise them into a sentence and say it. That act of bringing something to mind is retrieval.
And the other person is still standing there.
Conversation moves at a frankly unreasonable speed. Typical gaps between speakers are around a fifth of a second—shorter than the time needed to plan even a simple response from scratch. We often start preparing our answer while the other person is still talking. Levinson and Torreira, 2015
That makes “I understood everything, so why couldn’t I answer?” rather less mysterious.
Understanding matters. But conversation also asks you to retrieve and produce language quickly, on a schedule you don’t control.
In one experiment, people learning a small artificial language practised either producing it or comprehending it. The production group later did better on comprehension tests. A lab study isn’t a conversation in Zürich, but it suggests speaking can contribute to learning—not simply demonstrate what you’ve already learnt. Hopman and MacDonald, 2018
Recall is part of speaking: bringing something to mind makes it easier to find again later. Karpicke and Roediger, 2008 But language is not stored as ready-made sentences. You recall the pieces, then combine them to say what you mean.
You’ve learnt to say you want a coffee. Now say you don’t want one. Ask whether your friend wants one. Say you’d like one, but not right now.
The vocabulary stays familiar; the combination changes. That is what happens whenever you speak. The difficulty is not only remembering the words, but combining them while the conversation keeps moving.
So what would learning look like if we deliberately designed for that process?
It wouldn’t mean throwing a beginner into a conversation and hoping for the best. “Just talk” is fairly useless advice when you don’t yet have anything to work with.
It would mean structured speaking practice: start with language you know, respond to a prompt, then vary what you’re asked to say. A teacher—or a lesson designed around it—gives you time to respond, a cue when you’re stuck and feedback you can use on the next attempt.
Those awkward first attempts become part of the work: retrieval, sentence construction and speaking aloud. A pause is something you can work through, rather than a verdict on whether you’re any good at languages.
Speaking becomes part of learning, rather than something you hope to be ready for afterwards.
That’s what Echoa is built around. You hear an English cue, build the German sentence and say it aloud, and Sam, your teacher, responds to what you actually said. Fifteen-minute lessons, every answer out loud. Real conversations in 20 hours. A2 in weeks, not years.
We’d quite like the answer to arrive while we’re still in the café.