Read & Select
A single word appears. You have about five seconds to say whether it is real English. Half of them are invented — and the invented ones are built to look real.
- 5 sec on the clock, 15–18 per full test
- Feeds your Reading subscore
- Built for the current July 2025 DET format, including the new Interactive Speaking task
Read & Select is the fastest task on the test and the one candidates most often dismiss. A word appears; you press real word or not a word; the next one arrives. It runs fifteen to eighteen times in a full test, which makes it one of the highest-frequency items you will meet.
The difficulty is entirely in the fakes. They are not random letters — they are built from real English morphology, so flatuce, prandish and moliten all obey English spelling and stress patterns and all feel like words you might have seen. That is deliberate. A test that used xqmpz would measure nothing.
What it actually measures is the size and the firmness of your vocabulary: not just whether you have met a word, but whether you know you have met it.
What you actually see
One word, centred, large. Two buttons. A timer that is short enough that you will answer on instinct whether you intend to or not.
- A single word at a time — never a sentence, never any context.
- Roughly five seconds per word. Miss the window and it counts as unanswered.
- A mix of real English words and invented ones that follow English spelling rules.
- Real words range from very common to genuinely obscure, which is how the task discriminates at the top of the scale.
- No feedback between words. You never learn whether you were right.
Why the fake words are hard
Look at these six the way the task presents them — no context, five seconds each:
brindle · flatuce · sundry · prandish · quell · moliten
brindle, sundry and quell are real. flatuce, prandish and moliten are not.
Now notice what makes them hard. prandish carries the -ish suffix that gives you brandish and outlandish. moliten has the shape of molten crossed with a past participle in -en. flatuce borrows the Latinate tail of produce and flatulence. Each one activates a real word nearby, and that near-miss is what produces the false positive.
Meanwhile brindle — a real word, meaning a brown-grey coat pattern on an animal — is unfamiliar enough that people reject it. The task is squeezing you from both sides at once, and that is precisely how it separates a 100 from a 140.
Ten words, real or invented
The same ten from the example above, shuffled through one at a time. Answer on instinct.
How Read & Select is scored
This is the one task on the test where a wrong answer is known to be worse than no answer. Duolingo scores it so that selecting invented words as real counts against you — the task is explicitly designed to catch candidates who click real on everything.
That asymmetry drives the whole strategy. If you answered real to all eighteen words you would catch every real word and every fake, and the fakes would drag you down to something near the floor. Selecting only the words you actually recognise is not caution; it is the correct play.
Because the item pool spans very common to quite rare words, and because the test is adaptive, the words you are shown track your apparent level. Seeing words you have never met is not a bad sign — it usually means you have been doing well.
| What the scorer looks at | Why it matters |
|---|---|
| Real words correctly identified | The positive side of the ledger — straightforward vocabulary breadth. |
| Invented words wrongly accepted | Counts against you. This is the mechanism that makes clicking "real" on everything fail. |
| Real words wrongly rejected | A missed mark. Over-caution has a cost too, just a smaller one. |
| Rarity of the words you were served | Adaptive: recognising rarer words is stronger evidence of a high level. |
| Answering inside the window | An unanswered word scores nothing. At five seconds, hesitation is expensive. |
The mistakes that cost the most
Almost all lost marks on this task come from one of two opposite errors, and knowing which one is yours is most of the fix.
Do this
- Answer from recognition, not reasoning. If you have to construct an argument for why a word could exist, that is your answer.
- Trust a definite yes. If you can produce a meaning or a sentence for it instantly, mark it real even if it looks odd.
- Learn the common fake-word patterns — plausible suffixes bolted onto real stems is by far the most frequent shape.
- Keep moving. Five seconds means the pace is the point; a slow correct answer on one word costs you the next.
- Accept that you will meet real words you do not know. Rejecting them is the right call when you genuinely do not recognise them.
Not this
- Do not click "real word" as a default. The scoring is built specifically to punish this and it will collapse your score.
- Do not try to define a word to decide. There is not time, and familiarity is the signal being measured.
- Do not talk yourself into a word because a similar one exists — that similarity is the trap.
- Do not dwell on one you got wrong. There is no feedback, so the regret is speculation and it costs you the next word.
- Do not treat this as the place to deploy vocabulary you half-learned last week. Half-known words are exactly what produces false positives.
How to actually get better at Read & Select
Of all eleven current tasks this is the most improvable by ordinary reading, and the least improvable by cramming word lists — because a word you memorised yesterday is precisely a half-known word.
Read widely rather than studying lists
Vocabulary that is firm enough to survive a five-second judgement is vocabulary you have met repeatedly in context. Volume of reading beats any list.
Learn the shape of English morphology
Spend an hour on the common suffixes — -ish, -ous, -ate, -ence, -ile, -en — and how they attach. The fakes are made by attaching them wrongly, and once you can see the join you stop being fooled.
Practise at speed deliberately
Run rounds with a five-second cap from the start. Practising slowly trains a judgement you will not have time to make.
Track your error type, not your score
Count false positives (fakes you accepted) separately from false negatives (real words you rejected). They need opposite corrections and the total score hides which you have.
Read one thing outside your field each week
The rare real words come from everywhere. If everything you read is in one domain, the pool of words you firmly know stays narrow.
How much this one task moves your score
At fifteen to eighteen appearances, this is one of the most frequent tasks in a full test, and it feeds Reading — and so Literacy and Comprehension.
It is also the fastest task to stop losing points on, which is why it is worth auditing early. The correction for over-clicking real takes one session to understand and produces an immediate gain. The correction for a genuinely narrow vocabulary takes months of reading. Find out which one you have before you spend the months.
What it cannot do is carry a low score on its own. Vocabulary recognition is a necessary condition for a high DET score, not a sufficient one — you can recognise every word on the test and still score poorly if you cannot produce the language.
Read & Select: questions people ask
Do I lose marks for selecting a fake word as real?
Yes. Duolingo designs this task so that accepting invented words counts against you — it is the mechanism that stops a candidate scoring well by clicking "real word" on everything. Select only the words you genuinely recognise.
What if I do not know whether a word is real?
Mark it as not a word. Genuine recognition is what the task measures, and a word you cannot place is, for scoring purposes, a word you do not know. The risk of accepting a fake outweighs the missed mark on a real word you did not recognise.
How are the fake words made?
They follow English spelling and morphology — real prefixes and suffixes attached to plausible stems — so they look and sound like words that could exist. That is deliberate: it forces the task to measure actual familiarity rather than pattern-spotting.
How long do I get for each word?
About five seconds. It is the tightest timing on the test, and it is short by design: the task is measuring instant recognition, not deliberation.
Will studying a vocabulary list raise this score?
Less than you would hope, and it can backfire. Freshly memorised words are half-known, and half-known words are what produce false positives on the invented items. Broad reading over weeks builds the firm familiarity this task rewards.
Are the words shown one at a time or all together?
Individually — each word gets its own few-second window rather than sitting in a list you can scan and cherry-pick from. That single-item presentation is what makes the timing the tightest on the test, and it means there is no reading ahead and no coming back to reconsider.
Which subscores does Read & Select feed?
It feeds Reading, and through that the Literacy and Comprehension subscores — not Speaking or Writing, which come only from the tasks where you produce language. Its influence comes from frequency rather than any single item: see how the eight subscores are built.
Is there any audio in Read & Select?
None. You are reading printed words and deciding whether each is genuine English — nothing is spoken and nothing depends on your listening. The name misleads people into expecting audio, but this task is purely visual recognition.
How many words appear in Read & Select?
Fifteen to eighteen across a full test, which makes it the most frequent single item type you will meet. That frequency is why a systematic habit — like clicking "real word" when unsure — compounds into a large loss.
What proportion of the words are invented?
Roughly half in any given set, though the exact mix varies. Assuming a fixed ratio and answering to fit it is a mistake; judge each word on its own.
Do the words get harder as the test goes on?
They track your performance rather than the clock. Doing well brings rarer words; struggling brings more common ones. Meeting words you have never seen is usually a sign the engine has revised its estimate of you upward.
Are proper nouns or abbreviations ever used?
No. The items are ordinary English vocabulary — nouns, verbs, adjectives — so you never need to decide whether a place name or an acronym counts.
What happens if I run out of time on a word?
It counts as unanswered and scores nothing. At around five seconds per word the practical rule is to answer on first recognition; a slow correct answer on one word costs you the next.
Can I improve at Read & Select in a week?
You can fix a strategy error in a week — if you have been accepting invented words by default, correcting that produces an immediate gain. You cannot broaden a vocabulary in a week, and cramming lists actively hurts because half-known words are what the fakes catch.
Why did I reject a word that turned out to be real?
Because the item pool reaches into genuinely uncommon English — words like brindle or furlong. Rejecting a real word you do not recognise is still the correct decision under this scoring, since accepting fakes costs more. Broad reading over months is the only thing that shrinks that category.
Where Read & Select sits in a full test
No task on the DET is scored in isolation. The adaptive engine is building one picture of your English out of every response, so it helps to know how much of the hour this particular task accounts for and which of the others are drawing on the same ability.
The table below is the whole test. Your current page is highlighted.
| Question type | Time | How often | Subscores it feeds |
|---|---|---|---|
| Read & Complete | 3 minutes | 3–6 times | Reading, Writing |
| Read & Select — you are here | 5 seconds | 15–18 times | Reading |
| Fill in the Blanks | 20 seconds | 6–9 times | Reading, Writing |
| Listen & Type | 1 minute | 6–9 times | Listening, Writing |
| Speak About the Photo | 90 seconds | at least once | Speaking |
| Interactive Speaking | 35 seconds each | 6–7 questions | Speaking, Listening |
| Write About the Photo | 1 minute | at least twice | Writing |
| Interactive Reading | 7 minutes | 2 sets | Reading |
| Interactive Listening | 6 minutes | 2 sets | Listening, Speaking |
| Speaking Sample | 3 minutes | once | Speaking |
| Writing Sample | 5 minutes | once | Writing |
Two things are worth reading off that table. The first is that the tasks feeding Reading are not only this one — improvement transfers, so time spent here shows up elsewhere. The second is that the tasks taking the largest share of the hour are not the ones most people practise most.
Question types that use the same muscles
Read & Complete
A short passage appears with roughly every second word cut in half. You type the missing letters to restore each damaged word.
Read the guide → Reading + WritingFill in the Blanks
A sentence appears with one word partly deleted. The first letter or two are given; you type the rest.
Read the guide → ReadingInteractive Reading
A passage with five linked tasks: completing sentences, choosing what comes next, highlighting the part of the text that answers a question, identifying the main idea, and choosing a title.
Read the guide →Start a practice test
An hour under real conditions, all eight subscores, and feedback on every written and spoken answer.
Take a test