Listen and select plays spoken words one after another and asks which are real English. Because you never see the spelling, the Duolingo English Test is testing whether a word exists in your ear, not on the page — which is why candidates who read well can still find this task hard.
Repeating it internally often triggers recognition that listening alone does not. It takes half a second and converts a vague feeling into a decision.
If you can picture how the word would be written, you almost certainly know it. If no spelling comes at all, that is a strong signal it is invented.
English words have one strong syllable. If what you heard has a stress pattern you cannot place in any word you know, treat it with suspicion.
As with the written version, choosing every word tells the test nothing. The non-words you select count against the real words you found.
Give every word the same short window. Stalling on one word means the next two go past while you are still deciding.
Work through ship/sheep, full/fool, bad/bed, walk/work for ten minutes a day. Most listening errors at B2 are vowel-length errors, and they are fixable in a fortnight.
If all your listening practice comes from one source, an unfamiliar vowel will read as an invented word. Rotate between British, American, Irish and Indian English while you prepare.
Use headphones you have tested, at a level you have already chosen. Adjusting the volume during the task costs you words you would otherwise have known.
Not thisRejecting ‘sheep’ because you expected ‘ship’.
ThisHearing the vowel length and accepting both as real.
Long and short vowels carry meaning in English. If you cannot hear the difference, real words start to sound invented.
Not thisSelecting an invented word because it sounds like a word in your first language.
ThisAsking whether you have heard it in English.
Non-words are built to sound plausible, and plausible often means close to an international-sounding stem.
Not thisWaiting to be sure before selecting.
ThisDeciding within about two seconds.
The audio does not wait. Certainty that arrives after the next word has played is worth nothing.
Not thisSelecting every word once you start to panic.
ThisSelecting only what you recognise, even if that is few.
Wrong selections count. A panicked sweep produces a worse result than an honest, sparse one.
Not thisTesting with your laptop speakers the night before, then using headphones on the day.
ThisPractising and testing on the same equipment.
Your ear adapts to the sound you train on. Change it on test day and you lose words to the hardware.
Not thisAssuming a long, unusual-sounding word must be invented.
ThisRecognising that harder screens use lower-frequency words.
The test is adaptive, so unusual words are a sign of good performance, not of a trick.
40 original questions. The first 10 are open to everyone. Answers are checked in your browser and never sent to us, so you can share this page with a class as it is.
2Complete the real English word.
5Complete the real English word.
7Complete the real English word.
8A word plays that you cannot place at all. Describe your decision process in one sentence.
One good answer
I repeat it silently and try to picture a spelling; if no spelling and no stress pattern I recognise comes within about two seconds, I leave it unselected and give my full attention to the next word.
10Complete the real English word.
12Complete the real English word.
14Complete the real English word.
16Design a ten-minute daily exercise for the vowel-length errors that cause most problems here.
One good answer
Pick four pairs — ship/sheep, full/fool, bad/bed, live/leave — and listen to a recording of them in random order, writing down which I hear. Then record myself saying them and check whether the difference is audible in my own voice, since producing the contrast makes hearing it easier.
18Complete the real English word.
21Complete the real English word.
23Complete the real English word.
25Complete the real English word.
27Complete the real English word.
28Explain why a candidate who reads English well can still struggle with this task.
One good answer
Reading stores a word as a shape on the page, but this task only gives the sound. If I have met a word in books and never heard it spoken, I have no sound to match, so it arrives feeling invented even though I know it perfectly well in writing.
30Complete the real English word.
32Complete the real English word.
35Complete the real English word.
37Complete the real English word.
40Describe what you would do in the five minutes before this section begins.
One good answer
Set the volume once, using the headphones I have practised with, and leave it there. Then remind myself of the two-second rule and the fact that I am selecting only what I recognise, so that I do not drift into a defensive sweep when the first unfamiliar word arrives.
Note
Pick Two Activity is an independent tool. It is not affiliated with, endorsed by or sponsored by Cambridge University Press & Assessment, the British Council, IDP, ETS, Duolingo or any other exam board. Exam names are trademarks of their owners and are used here only to describe the exam you are preparing for. Every question on this page is original practice written by us: we do not provide, copy or reproduce official exam papers.
Because you get no spelling. A word you have only ever read has no sound attached to it in your memory, so it arrives feeling invented. Listening practice, not reading practice, fixes this.
Use whatever you have practised with, provided the current test rules allow it — check the official site, since equipment requirements can change. Consistency between practice and test day matters more than the equipment itself.
That is exactly what minimal pair practice prevents. Ten minutes a day on ship/sheep, full/fool and bad/bed removes a surprising share of these errors within two weeks.
Yes. They are built from real English syllables and follow English stress patterns, so you cannot reject them on sound alone. The judgement is recognition, not analysis.
There is no target. Select the words you recognise and no others. A sparse, honest set of selections tells the test more about your level than a full screen of guesses.