Don't learn all the colours on the same day

Open almost any beginner's language course and you will find the vocabulary arriving in tidy groups. Here is the page of colours. Here are the days of the week, Monday through Sunday. Here are the clothes — shirt, skirt, sweater, trousers, socks — and here, a chapter later, the fruits and the farm animals. It looks like the obviously sensible way to organise a language. Like things go with like things; learn them as a set.

It feels right, and it is one of the least efficient things you can do. When researchers have actually put it to the test, learning words in these neat categorical clusters comes out slower and more confusable than learning the very same words in a jumbled, unrelated heap. The tidiness that makes the page look organised is, from your memory's point of view, precisely the problem.

A finding that annoys textbooks

The clearest demonstration came from Thomas Tinkham, in a pair of experiments published in 1993 in the journal System. His design was deliberately plain. He gave English speakers a batch of invented words to learn, each paired with an English meaning. Half of those meanings belonged to a single category — a set of clothes, say, all sharing one overarching idea — while the other half were unrelated to each other, a random scatter of things with nothing in common.

If grouping helped, the categorical set should have been the easy half. It was the hard half. Words that shared a common superordinate concept — all of them "kinds of clothing," all of them "kinds of animal" — took more trials to learn and were recalled less reliably than the unrelated words. The set that looked organised on the page behaved, inside the learner, like a tangle.

The words were not harder because of anything about the words. They were harder because of the company they were forced to keep.

Why sameness breeds confusion

The reason has a long pedigree in the study of memory, and it runs exactly counter to intuition. We tend to assume that related things reinforce each other — that learning shirt gives skirt a leg up. In fact the opposite tends to happen, through a mechanism psychologists call interference.

Back in 1963, the psychologist Masanori Higa laid out the principle in a study with the unglamorous title "Interference effects of intralist word relationships in verbal learning." His finding, since replicated many times, was that when the items you are trying to learn at once are similar to one another — in meaning, in sound, in the associations they trigger — they compete. The more alike they are, the harder each one is to pin down. Similarity, which feels like an aid, is in learning terms a source of friction.

You can feel why once you picture what the mind has to do. To learn shirt and skirt and sweater together is to try to build several new associations that all point at nearly the same place — clothing, upper body, things you put on — and that resemble one another in form as well. The links get crossed. Later, reaching for one, you snag the other. This is the same reason a beginner who learned left and right in one breath spends years flinching before choosing between them, and why Tuesday and Thursday feel interchangeable long after the rest of the week has settled. Memory prizes distinctiveness; a category strips it away by design, handing you a set of items whose whole point is that they are alike.

Same words, different grouping

The obvious objection is that this simply argues against grouping words at all — and that would be a shame, because a good grouping is a powerful thing. Tinkham anticipated it. In a follow-up published in 1997 in Second Language Research, he pitted two different kinds of grouping against each other, and the result is the practical heart of the matter.

The losing kind he called semantic clustering: words bound together only by a shared category — a list of body parts, a list of vegetables. The winning kind he called thematic clustering: words bound together by a shared scene. Not frog, toad, newt, lizard (four amphibious near-neighbours, mutually confusable), but frog, green, pond, hop, croak — a cluster that belongs to one small story rather than one flat category. Words linked by a situation, Tinkham found, were learned more easily than unrelated words, not less. The scene gives each word a different job to do, and different jobs are easy to tell apart.

Two years earlier the effect had already survived an independent check. Robert Waring, in a 1997 replication pointedly titled "The negative effects of learning words in semantic sets," ran the experiment again with fresh materials and found the same main effect: presenting new words in semantic sets, rather than as unrelated groups, worked against the learner. When a result annoys the textbooks and reproduces, it is worth taking seriously.

Is the case closed?

It would be dishonest to pretend the whole field marches in step, because it doesn't. Reviews that gather up the studies find a genuine split: a majority report that semantic clustering hinders learning, but a sizeable minority find no difference, or even a small benefit. One recurring way of squaring the accounts is a distinction between input and output — semantic relatedness may make it slightly easier to produce a word you already hold, while making it harder to get the word in in the first place. The interference bites hardest at exactly the moment a beginner is most exposed: first contact, when the new forms are still fragile and easy to swap.

And the effect is not confined to the laboratory. In a 2008 classroom study published in System, İsmail Erten and Mustafa Tekin taught sixty young learners eighty carefully chosen words, some in semantically related sets and some in unrelated ones. The children who met the words in unrelated sets came out ahead — and the authors concluded that clustering by category may quietly hamper the very long-term retention it is meant to help. The safe reading of the whole literature is modest but clear: front-loading a category rarely helps, and often hurts, so there is little reason to keep doing it and a decent reason to stop.

What to do instead

None of this asks you to work harder. It asks you to group differently — and mostly, to stop grouping by category at all.

• Separate the confusable ones. If two new words share a meaning and a shape — shirt/skirt, left/right, the seven days — don't meet them together. Learn one, let it settle, and introduce its neighbour once the first is secure enough not to be knocked over.
• Learn by scene, not by set. Reach for a word inside the situation it lives in. Coffee, cup, hot, morning, spill hang together in a moment; coffee, tea, juice, milk, water are just a shelf of confusable drinks. The scene gives each word a distinct hook.
• Meet words in sentences. A word encountered in a real sentence arrives already distinct — it has a role, a neighbour, a job — which is a large part of why words learned in sentences outlast words learned on bare cards.
• Space the neighbours out. Interference is worst when similar items are crowded into the same session. Put time between them — reviewing each just as it begins to fade — and the crowding, and the confusion, ease off.

Where vocabla fits

This is one of the quiet reasons vocabla doesn't march you through the colours page and then the animals page. Words show up inside real sentences, one situation at a time, so each new word arrives with a role that sets it apart rather than a category that blurs it into its neighbours. Reviews are spaced so you reach for a word from memory at the moment it's starting to slip — which naturally pulls confusable pairs apart in time instead of stacking them into a single, muddling session. The point isn't to memorise a set. It's to meet each word where it's easy to tell apart, and then again later, until it stands on its own.

So the next time a course fans out a whole category in front of you — every colour, every item of clothing, all seven days at once — you can treat it as a reference page rather than a study plan. Learn the words the way you will actually use them: one at a time, inside a scene, spaced across the days. The tidy list was never for your memory. It was only ever for the page.