Duolingo English Test cheat sheet

Overall scores run from 10 to 160 in 5-point increments; the Literacy subscore is computed only from reading-plus-writing tasks, not from the whole test.

Read and Complete presents a passage with partially spelled missing words: the blank supplies the opening letter or letters, and the typed answer must match that letter count exactly.

Read and Select mixes genuine English words with invented ones; the invented items follow English spelling patterns, so recognized usage rather than spelling plausibility is the discriminator.

Read then Write requires a typed response to a short passage or prompt, scored on relevance to the prompt, grammatical accuracy, and vocabulary variety.

Difficulty adapts within the test: consecutive correct answers raise item difficulty, so the reported score reflects the hardest level reached rather than a raw percentage correct.

Approximate CEFR mapping: 10-55 (A1-A2), 60-95 (B1), 100-125 (B2), 130-150 (C1), 155-160 (C2).

Unanswered items are scored as incorrect, and per-item timers run continuously without pausing.

The overall score is not the arithmetic mean of the four subscores; it is derived from all item responses across sections.

Results are typically issued within about 48 hours and remain valid for two years; reports can be sent to an unlimited number of institutions at no extra cost.

Comprehension items are listening-and-reading based: Listen and Select and Listen and Type, plus reading items such as Read and Select.

Listen and Select plays a short sequence of words once each; only genuine English words are accepted, and the invented distractors are phonotactically legal English-shaped strings.

Listen and Type requires exact reproduction of the heard sentence, compared against the source for spelling, capitalization, and word order.

Partial credit applies in Listen and Type: correctly transcribed words earn credit even when the sentence is incomplete.

Audio is delivered at natural conversational speed with a mix of native accents, so word recognition must generalize beyond a single accent.

Most listening prompts play a fixed, limited number of times (commonly once) and there is no replay control.

Reading comprehension is assessed through short cloze and lexical-decision items rather than extended passages followed by comprehension questions.

A band score of roughly 100-125 corresponds to CEFR B2, the level most undergraduate admissions thresholds target.

The Conversation subscore derives from listening-and-speaking tasks: Listen, Then Speak and Interactive Listening.

Listen, Then Speak delivers a spoken prompt a limited number of times, after which the candidate records a timed spoken answer.

Interactive Listening presents a simulated dialogue in which the candidate selects the contextually appropriate reply and may add a recorded spoken follow-up.

Speech scoring weighs task relevance, fluency, grammatical accuracy, lexical range, and pronunciation; intelligibility matters, accent itself is not penalized.

Fluency measures penalize long silent pauses, filled pauses, and restarts, so steady simple speech can outscore hesitant complex speech.

Pragmatics are tested: formality, politeness, and turn-appropriate wording differ between peer contexts and authority contexts.

The recorded speaking sample (video interview) is unscored for the numeric result but travels with the score report and may be reviewed by institutions.

No human examiner is present, so misheard prompts cannot be repeated or clarified and must be inferred from context.

All subscores, including Conversation, are reported on the same 10-160 scale in 5-point steps.

The Production subscore comes from writing-and-speaking output tasks: Write About the Photo, Speak About the Photo, Read Aloud, and the writing sample.

Write About the Photo allows about one minute; naming visible objects, their positions, and ongoing actions outperforms speculation about unseen context.

Speak About the Photo gives a brief preparation window (roughly 30 seconds) followed by a recorded spoken description of the image.

Read Aloud is scored on exact reproduction: omitted, inserted, or substituted words and mispronounced words each reduce the item score.

Written responses are graded on task relevance, grammatical accuracy, vocabulary range, and coherence, with sentence variety as a scored feature.

The writing sample and video interview do not change the numeric score but are sent to institutions and screened for plagiarism and rule violations.

Repeating high-frequency words limits the lexical score; precise, less common vocabulary used in correct collocations raises it.

Discourse connectives such as 'however', 'therefore', and 'for instance' count toward coherence in both written and spoken responses.

Timers do not pause and unfinished text is submitted as-is, so accurate partial output outscores an incomplete ambitious answer.