Seven, plus or minus a word

Try this with a language you don't speak. Have someone say a longish word to you once — say the Finnish lentokenttä, airport — and then, a beat later, repeat it back. If you're a beginner, you probably can't. The middle dissolves; you're left clutching a first syllable and a vague shape. Now notice what happens with a phrase in your own language: "the airport was closed for three hours" is far more information, and yet you could hold it, turn it over, and hand it back without strain. The foreign word was shorter and it beat you anyway.

That isn't a defect in your memory. It's your memory running into a limit that everyone half-remembers from a psychology class — the famous "seven, plus or minus two" — and then, underneath it, a subtler truth that the famous number gets wrong. The gap between the phrase you can hold and the word you can't is the whole story of why new vocabulary is hard before it is easy.

The integer that persecuted a psychologist

In 1956 the psychologist George A. Miller published a paper in Psychological Review with one of the most quoted titles in all of science: "The Magical Number Seven, Plus or Minus Two." It opens, gloriously, with a complaint: "My problem is that I have been persecuted by an integer." Everywhere Miller looked — how many tones people could tell apart, how many digits they could echo back — the answer landed near seven. The span of immediate memory seemed to be about seven items, give or take a couple.

But the number is the part everybody keeps and the part Miller cared about least. His real argument was that the bottleneck isn't measured in information — it's measured in chunks. A chunk is a unit that has become meaningful enough to count as one thing. You can hold about seven of them whether each is a single digit or a whole memorised idea, which means the way past the limit isn't a bigger memory but better recoding: bundling small pieces into larger, familiar units so that each slot carries more. The magic was never the seven. It was the chunk.

The number was closer to four all along

Later work made the ceiling lower and the point sharper. When you strip away the tricks people quietly use to inflate their span — rehearsing under their breath, grouping digits into dates and phone numbers — the raw capacity is smaller than seven. In an influential 2001 review, the psychologist Nelson Cowan gathered the studies that controlled for those tricks and argued that the true limit of what we can hold in mind at once is closer to four chunks, not seven. Seven is what a practised mind achieves with chunking; four is roughly what's left when chunking is taken away.

And chunking, it turns out, is most of what expertise is. The cleanest demonstration is an old one about chess. In 1973, William Chase and Herbert Simon showed masters and novices a chessboard for a few seconds and asked them to reconstruct it. On real positions from real games, the masters were staggeringly better. But on boards with the pieces scattered at random, the masters lost their edge almost entirely — they were no better than beginners. The master's memory wasn't larger. It was that a real position, to a trained eye, isn't twenty-five separate pieces; it's a handful of familiar patterns. Randomise the board and the patterns vanish, and the master is back down to holding four-ish things like everyone else.

A beginner and an expert have the same small memory. The expert has simply learned to pack more into each of its few slots.

Why a new word isn't yet one thing

Here is where a foreign word walks straight into the trap. To someone who already speaks the language, lentokenttä is one chunk — one word, one meaning, one shape the mouth already knows. To you, on first hearing, it is not one thing at all. It's a loose string of unfamiliar syllables, none of them a unit yet, each competing for one of those precious four slots. The word is short on paper and enormous in your head.

The mechanism has a name. Verbal material is held in what Alan Baddeley called the phonological loop — a kind of inner voice that rehearses sound for a couple of seconds before it fades. Its capacity is roughly however much you can say in about two seconds, which is why memory span is reliably better for short words than long ones: the loop empties in real time, and a long, unfamiliar form fills it before you can circle back to refresh the start. A word you can't yet pronounce fluently takes longer to rehearse, so it decays before you finish holding it. That's the crumbling middle of lentokenttä, explained.

None of this is idle theory for language learners. Across a body of work summed up by Baddeley, Susan Gathercole and Costanza Papagno in a 1998 paper pointedly titled "The phonological loop as a language learning device," the capacity of phonological short-term memory turns out to predict how quickly people — children and adults alike — pick up new vocabulary. The ability to hold an unfamiliar sound-pattern in mind long enough to do anything with it is not a side detail of word learning. It may be the front door.

The catch that keeps it honest

So should you train your memory span, drill nonsense syllables, try to widen the four slots to five? No — and this is the part the productivity version always skips. Working memory is a workbench, not a warehouse. Its job is to hold a few things while you do something to them; it is not where anything is stored. You can chunk a word perfectly, hold it, repeat it back flawlessly, and still not know it, because knowing a word means it has moved off the bench and into long-term memory — and that move is a different process entirely, made of repeated encounters spaced out over days, not one heroic act of holding.

There's a second honest catch, the one that shadows every technique on this site. Getting a foreign word to fit in working memory — turning its sound into a single smooth chunk — is real progress, but it's progress on the form, not the meaning. A word you can now say cleanly is a word you can finally start attaching a meaning to; it isn't a word you've learned. The loop hands you the shape. Wiring that shape to what it means, and keeping it, is still the slow work.

Working with the bottleneck instead of against it

Once you stop trying to beat the four-slot limit and start respecting it, a lot of good advice falls out of the same idea.

• Take words a handful at a time. A page of fifty new words isn't ambitious, it's a jam: forty-six of them never get a slot on the bench. A few words, met properly, actually get processed. The limit is telling you what a real study batch looks like.
• Say it to loop it. The phonological loop runs on your own voice, so a word you only read silently barely enters it. Pronouncing a new word — as we argued in saying it out loud — is partly just feeding the loop something it can hold long enough to work on.
• Learn the bigger chunk on purpose. The way past the limit is recoding, so learn words in the phrases they travel in and inside their families. "Strong coffee" or a familiar sentence frame becomes one chunk, not three, and one chunk is something you can actually hold.
• Don't crowd similar words together. A batch of near-identical words doesn't chunk cleanly — they blur into one another and compete for the same slot, which is exactly why learning a whole category at once backfires.
• Then let the shape settle before you judge it. A word that crumbles today is a word your loop can't yet chunk. Meet it a few more times and the syllables fuse; the crumbling stops on its own. That isn't your memory improving — it's the word becoming one thing.

Where vocabla fits

vocabla is built for a bench with four slots, not a warehouse with infinite shelves. New words arrive a small handful at a time rather than as a wall of a list, so each one gets room to be processed instead of shoved off the edge. They come inside sentences — which is to say, inside a ready-made chunk, a frame your mind can hold as one unit while the new word rides along inside it. And because the words return on a schedule instead of all at once, the form has time to fuse into a single smooth shape and then to move, quietly, off the bench and into the place where words actually live.

The magical number was never seven, and it isn't really four either. It's however many things you've made familiar enough to count as one. Every word you learn is one more unit that used to cost you the whole bench and now costs you nothing — until, one day, a stranger says lentokenttä and you hand it right back.