How to learn English words: how many you need and how to make them stay

How many English words a learner needs, counted in word families

People who search how to learn english words often expect a small number. Paul Nation of Victoria University of Wellington answered the question in a 2006 paper in the Canadian Modern Language Review. He counted word families: a base word with its inflections and common derived forms, so nation, national and nationally count once. Then he asked what vocabulary it takes to read a novel, watch a film or follow a conversation without help.

His answer is 8,000 to 9,000 word families for written English and 6,000 to 7,000 for spoken English, if a reader must know 98% of the running words. For scale, the paper cites estimates of 114,000 and 88,500 families in the whole language and puts a well-educated native speaker at around 20,000. A learner needs a few thousand of them, taken in the right order.

Two cautions apply. Word families are generous units: Nation notes that most learners have smaller families than native speakers, so someone who knows act may still stumble over activity. The figures also describe understanding, not speaking. You can follow far more words than you can produce, so the goals for reading and for talking differ.

Frequency first: the thousand English words that carry most of a text

Words are not equally useful. In Nation’s tables the first 1,000 word families cover about 78–81% of the running words in written English and 81–84% in speech. The second thousand adds 8–9% in writing and 5–6% in speech. From the sixth thousand on, each new thousand adds well under one per cent.

Coverage decides whether a page is readable. In a study by Hu and Nation summarised in the paper, readers who knew 80% of the words in a fiction text did not reach adequate comprehension, and at 95% only a small minority did. The paper translates 95% into one unknown word in every two lines of print, or about seven a minute in speech. At 98% it is one word in fifty.

The rule follows: learn the most frequent words first, because each of them pays back on every page. A themed list of zoo animals can wait until you need it. Frequency lists built from large collections of real text, called corpora, rank words by how often and how widely they occur. Working down such a list buys the most understanding for each hour.

Vocabulary size at each stage: graded readers, conversation and novels

The paper also gives stages. A graded-reader edition of The Picture of Dorian Gray, kept within a 1,000-word list, needs the first 2,000 families plus proper nouns for 96.75% coverage and 3,000 for 98.86%. So two or three thousand families open simplified books without a dictionary. In Nation’s samples of talk-back radio and friendly conversation, 3,000 families plus proper nouns gave over 95% coverage.

The Turn of the Screw and The Great Gatsby, written for native readers, reached 98% at 7,000 and 8,000 families plus proper nouns, and five newspaper corpora needed at least 8,000. Even the script of Shrek needed 7,000, because every story brings rare words of its own. Ogre, swamp and donkey run through it, yet beyond the fifth thousand only eight words occur in both Shrek and Toy Story.

Collocations and chunks: English words learnt in company

A word on a list is only half known. Wikipedia’s article on collocation defines it as words that co-occur more often than chance would predict, such as crystal clear, nuclear family or make with decision. A grammatically correct sentence, the article adds, still sounds awkward when these preferences are broken. Someone who knows make, do and decision separately can still say do a decision.

Nation’s paper points the same way. Transparent phrases matter little for reading and listening, it says, but must be learnt for speaking and writing, where each word takes its own collocates and grammar patterns. The English Vocabulary Profile, part of the English Profile Programme, holds words, phrases, phrasal verbs and idioms. An article in the English Profile Journal explains that it orders each meaning of a word by its CEFR level.

Phrases multiply as the level rises. At the time of that 2012 study, the profile listed 4 phrasal verbs at A1, 27 at A2, 145 at B1 and 263 at B2. So learn each word with its company: make a decision, depend on, heavy rain. A phrase is stored as one piece and comes out with the grammar already inside.

The forgetting curve and the spacing that answers it

Between 1880 and 1885 Hermann Ebbinghaus tested his own memory of nonsense syllables such as WID and ZOF. Wikipedia’s article on the forgetting curve sums up the result: people tend to halve their memory of new material within days or weeks unless they review it. It also reports evidence that a single review works best after about 10–20% of the time until the information is needed.

The answer is spaced repetition, which Wikipedia describes as an evidence-based technique usually done with flashcards. In the 1970s the German science journalist Sebastian Leitner proposed a learning box split into sections. A recalled card moves on to the next section and a missed one goes back to the first. Failing words come round often, while known ones wait longer.

Software now calculates the gaps. The open-source Free Spaced Repetition Scheduler, FSRS, models memory with difficulty, stability and retrievability, the probability of recalling an item at a given moment. Its guide defines stability as the days it takes that probability to fall from 100% to 90%. As of 20 October 2026, it calls a desired retention between 70% and 97% reasonable and warns that a higher target means more daily reviews.

Recall, not rereading: flashcards that make English words stick

How you review matters as much as when. In 2008 Jeffrey Karpicke and Henry Roediger reported in Science an experiment in which students learnt 40 Swahili–English word pairs, such as mashua and boat, as flashcards. Once recalled, a pair either stayed in the tests or was dropped from them. A week later, pairs that kept being tested were recalled at about 80%, and dropped ones at 36% and 33%, whether or not they were restudied.

The students did not see it coming. In every group they predicted they would remember about half of the pairs. Rereading a list sets the same trap, since familiar words feel known. Only a test from memory, in which you produce the word before you see it, shows what will still be there next week.

Different cards train different sides of a word. A meaning card shows the English and asks for its sense. A reverse card shows the meaning and asks you to type or say the English, which is closer to speaking. A gap card hides the word in a sentence, and a sound card plays it without the spelling.

A 30-minute English vocabulary routine and the habits that waste it

Half an hour a day works if it is spent in the right order. Start with ten minutes of due reviews, because fading words come before new ones. Then take ten to fifteen new words or phrases from a frequency list or a text you are reading, each with a sentence, a sound and its partners. In the last minutes, use them: say three sentences aloud and write two about your day.

Three habits waste that half hour. The first is single words with translations and no context: they come back on a card but come out wrong in a sentence. The second is recognition only. Wikipedia’s article on the testing effect notes that short-answer tests and essays bring larger gains than multiple-choice ones, and a card that only asks for the meaning trains the easier half. The third is silence, since a word never said aloud rarely arrives in time.

Building English vocabulary alone or with the English in 213 Days course

Much of this works without a course. This week, pick 30 frequent words or phrases you understand but do not use, make cards of several kinds, review the due ones daily and say each phrase in a sentence about your life. On Sunday, test yourself from meaning to English and note what fails. That record is next week’s material and shows how to learn english words at a pace you can keep.

A structured course adds order and volume. English in 213 Days, the online intensive of VITON13 SCHOOL, has 213 levels, one a day, at 4–6 hours a day. Each level brings 13 new words with sound, IPA transcription and an example, and the day’s bank adds 52 words and phrases in four sets of 13: frequent words, collocations, things people say, and everyday verbs, phrasal verbs or idioms by band. That makes 13,845 in all, each joining a deck of FSRS cards: meaning, reverse, gap, pairs and sound.

The levels follow the CEFR bands from A1 to C1 and end on a road to C2; the course promises no level, since only an exam confirms one. In the English edition, explanations come in Russian, French, Spanish, German or Portuguese. The first three levels open without payment or sign-in, and a free placement test shows where to start. Either way the method stays: frequent words first, in phrases, recalled on a schedule.