The short answer
You get an overall score from 10 to 160, plus subscores. The four integrated subscores — Literacy, Comprehension, Conversation and Production — each pair two skills together, and they are what most institutions read when they look past the headline number. The test is adaptive: it gets harder when you answer well, so the difficulty you sustain matters as much as the number of correct answers.
Why the test gets harder as you go
The DET is computer-adaptive. Early answers shape what comes next: answer well and the questions get harder, struggle and they get easier. The score reflects the level of difficulty you could sustain, which is why two candidates who answered the same number of questions correctly can score very differently.
This has a practical consequence that surprises people. Questions getting hard is a good sign. Candidates routinely leave the test convinced they failed because the second half felt brutal, when the difficulty is exactly what a strong performance produces. Do not let it rattle you mid-test.
It also means no practice platform, including this one, can perfectly reproduce a real test score. Practice questions are fixed in difficulty; the real test calibrates to you. Treat any practice estimate — ours included — as a guide to readiness, not a prediction.
Not sure where you are starting from? Four questions, about two minutes, no sign-up — and you get an estimated score out of 160 to measure the rest of this against.
The subscores, and which ones matter
Underneath the overall score sit four skills: Reading, Writing, Listening and Speaking. Those are then paired into the four integrated subscores that appear on your certificate. The pairing is deliberate — each describes something you can actually do, rather than which paper you sat.
| Subscore | Built from | What it describes |
|---|---|---|
| Literacy | Reading + Writing | Understanding and producing written language. |
| Comprehension | Reading + Listening | Understanding written and spoken language. |
| Conversation | Speaking + Listening | Understanding and producing spoken language. |
| Production | Speaking + Writing | Producing language yourself, written and spoken. |
Institutions that set a subscore minimum usually pick the one that matches what the course demands. A taught master’s with heavy seminar participation may set a Conversation minimum; a research degree may care about Literacy. Check this before you prepare, because it changes what you should practise.
Why averaging is a trap
The overall score is an average, and an average hides a weak skill. A candidate strong in reading and listening can reach a respectable overall score while Production stays well below it — and Production is the subscore most often set as a minimum, because producing language is harder than recognising it.
If you are short on time, the highest-value thing you can do is find your weakest producing skill and work on that, rather than adding marks to a skill that is already above your target.
How writing and speaking are judged
Open-ended answers are scored on the qualities that make language work: grammatical range and accuracy, vocabulary, how well the response addresses the task, and for speaking, how fluently you kept going. Spelling and punctuation matter in writing; in speaking they do not exist, and a transcript full of missing commas tells you nothing about your score.
- Length is a floor, not a target. Answers below the expected length are penalised, but padding a short answer with filler does not raise the score and usually lowers accuracy.
- Keep speaking. Long silences cost more than a sentence that starts badly and recovers. Fluency is measured in what you produced, not in what you meant to say.
- Range is rewarded. A correct simple sentence is safe; a correct complex one shows more. Attempting range and getting it wrong is the expensive version, so build the structures you can control.
Where practice fits
Because the test is adaptive and the subscores are paired, the useful question is not “what did I score” but “which skill is dragging the pair down”. That is what practice can tell you reliably, and it is what the individual question types are best used for — each one isolates a skill in a way the real test deliberately does not.