You don't learn words. You learn chunks.

Picture someone ordering a coffee in a language they're learning. You can hear the assembly happening in real time. I... would like... a... uhh... coffee... please. The pauses are exactly the seconds where each word is being retrieved and placed.

Now picture a fluent speaker doing the same thing. I'll have a coffee, please.* Same meaning. Five times faster. The fluent speaker did not retrieve five separate words. They reached for one chunk — *I'll have a [thing], please — and dropped the thing into the slot.

This is not stylistic difference. This is the actual mechanism of fluency. And it is the single biggest reason most adult learners stay slow even after years of study.

Your brain stores phrases as one piece

Linguists have a name for these prefabricated phrases: formulaic language. Nice to meet you. As far as I know. To be honest. What I mean is. I'd rather not. Long story short. Each of these is multiple words. Your brain stores and retrieves each as a single chunk.

Estimates of how much of fluent native speech is made of such chunks vary, but they cluster in the same surprising range: somewhere between half and three-quarters. Most of what proficient speakers say is not freshly composed. It is recombined from a library of ready-made expressions, with new content slotted in.

This solves a problem that pure word-by-word learning never could. Speaking in real time is cognitively expensive. There isn't enough room in working memory to design every sentence from scratch while also tracking what your conversation partner is saying, what you want to say next, and where the waiter is. Chunks collapse that cost. They give your brain pre-built pieces it can deploy without thinking.

When a fluent speaker says "to be honest with you, I'm not sure," they are running roughly two cognitive operations. When a learner says the same thing word by word, they are running ten. No amount of vocabulary memorisation closes that gap. You can't word-by-word your way to chunk-speed.

Why most apps don't teach this

Most language apps teach vocabulary as isolated lemmas. A noun appears as just the noun. A verb appears as the infinitive. You learn coffee*, you learn *want*, you learn *please*, you learn *to have. Then, somehow, on your own, in real time, with a barista waiting, you are supposed to fuse them into a sentence.

It rarely works the way the syllabus implies.

The reason apps teach this way is that isolated words are easy to put in a flashcard. They have a clean question and a clean answer. The whole stack of card-based learning is built around that simplicity. The problem is that the simplicity doesn't match how people actually speak.

What Jaimy does instead

Jaimy is built around teaching chunks first and analytic vocabulary second.

When you arrive at a new level, the first things you encounter are not lists of words. They are the survival expressions for that level — the phrases you will reuse in dozens of conversations.

At A1: I'd like..., Where is...?, How much is it?, Can you repeat that?, I don't understand, see you later, nice to meet you. Roughly a hundred of these, drilled in context until they roll off the tongue.

At B1: I was thinking about..., the thing is..., it depends on..., I'd rather we..., if you don't mind..., could you tell me whether...? Now you can express opinion, qualification, polite request — without composing them from scratch.

At C1: I take your point, but..., on the face of it..., it strikes me as..., let me put it this way..., what you're really saying is... The discourse markers and softeners that make speech sound fluent rather than translated.

These chunks get reused across sessions. They turn up again in week three, week five, week ten, until they're as automatic as your own name. When you finally hit them in a real conversation, they don't have to be assembled. They're already there.

Vocabulary still matters — just not first

This is not an argument against learning words. You absolutely need a vocabulary, and Jaimy teaches plenty of them. The argument is about sequence and framing.

Words get learned as they appear inside chunks. You learn "nice to meet you*" first, as a unit, and the meaning of *meet* comes along for the ride. Later, when *meet* shows up in *meeting tomorrow* or *meet me at, you already have an anchor for it.

This is also how children acquire their first language. They don't learn the word "go" and then later combine it into "let's go." They learn "let's go" as one undifferentiated thing for months, and the analysis comes later, almost imperceptibly, from exposure to variations.

We're not children, and we have advantages they don't (we can ask questions, we can read, we can use translation as scaffolding). But the chunk-first sequence is one place where the children-acquisition path turns out to be closer to optimal than the textbook path.

The test you can do this week

Listen to yourself, in the language you're learning, doing a basic exchange — ordering, greeting, asking for directions. Count the pauses.

Then listen for the moments where you reached for a phrase that came out as one piece, with no internal hesitation. Those are your chunks. They're what fluent speech is made of.

The work, between now and proficiency, is to grow that set. Not the list of words you know — the list of expressions you can deploy without thinking.

Jaimy's job is to put those expressions in your mouth, in context, again and again, until they're yours.