Work from the word outwards: decide what part of speech it is, see what its own clause says about it, look at the sentences around it, then guess and test the guess by putting it back into the sentence. Expect misses: guessing works best when nearly every other word is familiar, and even good contexts rarely give an exact meaning.
What are the steps for guessing a word?
The best-known method comes from a 1980 paper by David Clarke and Paul Nation. It has four steps, and they stress that each one matters more than the exact order:
- Part of speech. Look at the word and its neighbours. Is it a noun, a verb, an adjective?
- Immediate context. What does the clause or sentence say about it? What does it do, or what is done to it?
- Wider context. What do the sentences before and after add? Linking words such as but, because and so often give the relationship away.
- Guess, then check. Is your guess the same part of speech? Do any prefixes or roots support it? Does the sentence still make sense with your guess put in its place?
Notice that word parts come last. Clarke and Nation argue that splitting a word into prefix and root tempts you to guess before you have read the context, and that although 25–30% of the vocabulary above the 3,000-word level can be broken into parts, only a small share of those words can easily be guessed that way.
Here is the method on a line from a free A2 chapter, with carpeta as the unknown word:
Llevaba una carpeta llena de partituras y una sonrisa nerviosa.
Una before it and llena (full) after it make carpeta a noun. Llevaba says it is something you carry. Llena de partituras says it holds sheet music, and the story is about a choir. A folder fits. Now the check: carpeta looks like the English carpet, but nobody carries a carpet full of music, so the look-alike fails the substitution test and the context wins.
How often are guesses right?
Less often than learners hope. Four studies measured it in different ways:
| Study | What learners did | What happened |
|---|---|---|
| Liu & Nation (1985) | Fifty-nine teachers of English, from native speakers to weak readers, guessed nonsense words standing in for rarer words, at 1 unknown word in 10 or 1 in 25 | Individual answers were correct 23–38% of the time. Lower density helped on one set of passages but not the other. Verbs were easiest, adjectives hardest. Pooling their guesses, a mixed group got 85% or more. |
| Nassaji (2003) | Twenty-one intermediate learners of English thought aloud while inferring unknown words in a text | Of 199 inferences, 25.6% were successful, 18.6% partly successful and 55.8% wrong. |
| Bensoussan & Laufer (1984) | Sixty Israeli university students translated 70 words alone, then again inside a text | Context helped in only 13% of responses. Stronger students knew more words but did not use context better. |
| Webb (2008) | Fifty Japanese university students met 10 disguised words, each in three sentences that gave more or less information | With the more informative sentences, students recalled the meaning of 1.31 words out of 10 on average; with the less informative ones, 0.13. |
Webb's numbers deserve a second look. Better contexts made a clear difference to recognising the meaning (6.77 against 4.38 out of 10 on a multiple-choice test), but even then, three good sentences rarely let anyone produce the meaning unaided. Context quality also made no significant difference to learning the spelling; that came from the number of meetings.
Two findings soften the picture. Liu and Nation's teachers guessed far better together than alone, which is why they recommend practising the strategy as a class before going solo. And in their economics passages, words were easier to guess when fewer of the surrounding words were unknown, which links guessing to vocabulary coverage: the fewer gaps around a word, the more clues survive to fill it.
Why do some guesses go wrong?
Some contexts point the wrong way. Webb cites earlier work by Beck, McKeown and McCaslin suggesting that many natural contexts are deceptive and lead readers to an incorrect meaning. In Hulstijn's 1992 experiments, readers left to infer without any clue often settled on plausible but wrong meanings, and then learned them.
Look-alike words are the classic trap, because the guess feels confirmed before the context is consulted. Bensoussan and Laufer found that students often applied a meaning they had already decided on, regardless of the text. This line from a free A2 chapter shows how a sentence can fail to warn you:
Pilar Ortega abrió la librería Cuatro Vientos y el gato color naranja siguió dormido en la silla.
Words with several meanings cause the same trouble from the other side: you know one meaning, and it happens to be the wrong one here. Our page on tap-to-translate or a dictionary shows two examples from the same series.
When should you guess, and when should you look it up?
- Guess and move on when the word isn't needed to follow the story, or when your guess makes good sense and you are reading for pleasure.
- Guess, then check when the word seems important. In Hulstijn's experiments, meanings readers worked out themselves were remembered better than meanings handed to them, so a guess followed by a quick tap gets you the effort and the right answer.
- Look it up straight away when the sentence makes no sense without it, when the word looks like one in your language, or when you meet it for the third time.
- Move down a level if you are guessing in every line. With too many gaps, there is not enough context left to guess from; our page on choosing your reading level has a quick test.
Read a free chapter and guess three words before you tap them. Count how many you got right.
Read free in your browserFrequently asked questions
Is guessing from context a good way to learn new words?
It is a good way to keep reading, but a slow way to learn exact meanings. In Webb's study, three informative sentences let learners recall the meaning of only about one target word in ten. Meeting the word many times, or checking and saving it, does the rest.
Which kinds of words are easiest to guess?
In Liu and Nation's study, verbs were the easiest, followed by nouns, with adverbs and adjectives the hardest. Nouns and verbs are also the most common kinds of unknown word, which helps.
What are false friends, and how do I catch them?
Words that look like a word in your language but mean something else, such as Spanish librería (bookshop) or carpeta (folder). Catch them with the substitution check: put the look-alike meaning back into the sentence and see if it really fits.
Should complete beginners try to guess words?
Only in easy text. Even in Liu and Nation's group of English teachers, individual guesses were right less than 40% of the time, and Nassaji's intermediate learners were fully right about a quarter of the time. At the start, tap freely and save guessing for sentences you otherwise understand.
Does guessing help me remember a word?
It can. Hulstijn found that meanings learners inferred themselves were remembered better than meanings given to them, but unguided guessers also learned wrong meanings. Guess first, then confirm.