Here's a number worth sitting with. Take the two thousand most common words in English, and they'll account for somewhere around eighty percent of the words on almost any page you read. Two thousand. Out of a language with hundreds of thousands of entries in its biggest dictionaries. You could ignore the other ninety-eight percent of the dictionary and still understand four out of every five words in a newspaper, a novel, a text from a friend.
That's not a coincidence, and it's not a quirk of English. It's a law — a mathematical one — and once you see it, it changes how you'd ever spend a minute of vocabulary study again. This is the section where the course gets ruthless about a question every learner secretly worries about: of all the words out there, which ones actually deserve your time?
That last chapter ended on who does the work — you or the source — when a word gets connected into your web of meaning. But before any of that work can happen, you have to decide which word goes on the board at all. And the research has a strong opinion about it.
So let's start with the law. It's called Zipf's law, after George Kingsley Zipf, a Harvard linguist who noticed it back in the 1930s. The idea is almost eerily simple. Rank every word in a language by how often it shows up. The most common word — in English, that's "the" — appears roughly twice as often as the second most common word, three times as often as the third, ten times as often as the tenth, and so on down the line. There's a 2014 review of this in the journal Psychonomic Bulletin and Review by the cognitive scientist Steven Piantadosi, and he calls Zipf's law one of the most puzzling facts about human language — precisely because it's so mathematically clean. Words didn't agree to behave this way. They just do, in every language ever measured, including extinct ones nobody can even translate.
Here's the plain-English version of what that means. A tiny handful of words do enormous amounts of work, and a vast crowd of words each do almost none. Piantadosi puts it nicely with examples: you've got your "a," "the," "I" doing the heavy lifting, and then way out in the long tail you've got words like "accordion," "catamaran," and "ravioli" — words you might go weeks without encountering. Most of the dictionary is catamarans and raviolis. The everyday business of understanding language runs on a small, busy core.
Picture a busy restaurant kitchen for a second. There are maybe a thousand ingredients in the whole pantry, but ninety percent of every dish that goes out the door uses the same forty or fifty: salt, oil, onion, garlic, flour, butter. The saffron and the truffle oil matter — for specific dishes — but a cook who mastered the forty workhorses first could feed the whole room. A cook who started by memorizing the rare spices would starve the dinner rush. Vocabulary works exactly the same way. There's a core that shows up everywhere, and a long tail that shows up almost nowhere.
So here's the obvious conclusion, and it really is this obvious: learn the high-frequency words first. If two thousand words get you eighty percent coverage, those two thousand are the highest-return study you can possibly do. Every one of them pays off constantly, in nearly everything you read or hear. A rare word, by definition, pays off rarely. This is the central case Paul Nation makes in his book Learning Vocabulary in Another Language — which is, by wide agreement, the standard reference in the field. Nation's whole argument is that a systematic approach to vocabulary means getting the best return for your learning effort, and the single biggest lever on return is frequency. Spend your limited time where the words actually live.
Now here's where it gets better than you'd expect — and this is the part that ties back to everything earlier in the course. Frequent words don't just pay off more once you know them. They're also easier to learn in the first place, and they get learned almost for free.
Think about why. Throughout this course we've kept coming back to two of the four levers — spacing and retrieval. A word sticks when you encounter it again and again, spread out over time, each encounter pulling it back out of memory. Well — what is a high-frequency word, if not a word that automatically gives you spaced, repeated encounters whether you study it or not? You don't have to schedule reviews for "because" or "important." Life schedules them for you. You meet the word today in an email, tomorrow in a podcast, the day after in a sign on the wall. The world becomes your spaced-repetition system, running in the background, for the words that matter most.
There's direct evidence for this in the incidental-learning research — the work on picking words up from reading, listening, and video without trying to. A 2022 study published in Frontiers in Psychology, led by researchers studying Spanish learners, had university students watch a video seeded with new words. Some of those words appeared once. Some appeared four times. Some appeared eight times. And the finding was exactly what you'd guess: the more often a word showed up, the more likely people were to recall it and recognize it afterward. The researchers put it plainly — there's a positive correlation between vocabulary growth and frequency of occurrence. In other words, the words you bump into most are the words that lodge themselves in your head with the least deliberate effort.
So if someone stopped you right here and asked why high-frequency words are doubly worth it — what would you say? Two reasons, stacked. They pay off more often once you know them, and they cost less to learn, because their own frequency keeps feeding you the repeated, spaced encounters that build memory. The rare word is the opposite on both counts. It rarely pays, and you'll almost never run into it often enough to learn it by accident.
That sets up a real tension, though, and it's worth being honest about it — because the "just learn the frequent words" advice can be taken too far. Here's the catch. The most frequent words in any language are gloriously useless on their own. "The," "of," "and," "to," "a" — the very top of the Zipf list is function words, the grammatical glue. You almost certainly already know those. So when people say "learn high-frequency words first," they don't mean start at rank one and march down. They mean target the high-frequency words you don't yet know — which, for most learners, is the band of common, meaning-carrying words that sit just past the absolute top of the list.
And there's a second wrinkle, the one that actually matters for how you steer your own study. General frequency lists are built from general language — newspapers, novels, everyday conversation. But your goals might not be general. If you're a nurse, "myocardial" is a rare word in the language at large and a daily word in your world. If you're learning Spanish to talk about cooking, the kitchen vocabulary that's rare in a national corpus is exactly what you need. This is the balance every serious learner has to strike: a foundation of general high-frequency words, plus a deliberately chosen set of specialized words your actual life requires.
Nation's framework handles this cleanly. The idea is roughly that you build the general high-frequency base — the few thousand words that show up everywhere — and then layer on top of it the specialized vocabulary of your field or interest, which is its own little high-frequency list inside your particular world. Both are frequency arguments. One uses the frequency of the language; the other uses the frequency of your life. The rare words to genuinely deprioritize are the ones that are rare in both — rare in the language and irrelevant to your goals. Those are the catamarans you can safely leave in the dictionary.
This is where the popular instinct goes wrong, and it's worth naming the disagreement directly. There's a romantic view of vocabulary — call it the "collect the beautiful rare words" school — where the goal is to hoard sesquipedalian gems like "defenestration" and "petrichor." It's charming, and it makes for fun social media posts. But as a strategy for actually becoming fluent or a stronger reader, it's backwards. Nation's frequency-first position has the weight of the evidence behind it, and the Zipf data is the reason why: those gorgeous rare words appear so seldom that no amount of memorizing them moves the needle on how much language you can actually handle. Learning "defenestration" feels productive. Learning the two hundred common words you keep half-knowing is productive. The feeling and the payoff point in opposite directions — and that gap is the whole trap.
So how do you actually find these words? The good news is you don't have to build the list yourself. Frequency-ranked word lists already exist, drawn from large collections of real text that linguists call corpora — basically, giant databases of actual written and spoken language, counted up. Nation himself is behind some of the most widely used ones in English-language teaching, organized into frequency bands: the first thousand most common words, the second thousand, and so on. The practical move is simple. Get a frequency list for your language, work through it more or less in order, and skip the words you already know solidly — your study time goes to the highest-frequency words you haven't yet locked in. Then add your specialized set on top. That's the whole prioritization scheme.
Strip away the detail, and three things are doing the real work in this section. Language is wildly lopsided — a small core of words covers most of what you'll ever encounter, and that's not opinion, it's Zipf's law. Because frequent words keep showing up on their own, they're both the most useful to know and the cheapest to learn, since the world supplies the spaced repetition for free. And "high-frequency" has to be read against your own goals, not just the dictionary — the general core plus the specialized words your life actually demands.
Here's the line worth carrying out of this chapter: the rarest words feel like the prize, but the common ones are where fluency is actually built. Be ruthless about it. Your study minutes are scarce, and the words aren't all worth the same.
Which means the next problem is purely mechanical. Once you've picked the right words to learn, something has to decide when to put each one back in front of you — and it turns out a piece of software can run that schedule better than you ever could by hand.