Task 1·~10 sentences · 10s response window each
Listen and Repeat
Hear a sentence read aloud once, then repeat it as accurately as you can — same words, same intonation.
What it tests: Pronunciation accuracy, intonation, rhythm, and your ability to hold a full sentence in working memory.
Tips that move your score
- Mimic the speaker's stress and rhythm, not just the words. Graders score intonation separately — a flat reading of correct words scores lower than expressive delivery.
- Listen for the END of the sentence as carefully as the start. Most repeaters drop the last 2-3 words because they're already speaking. Stay alert through the full sentence.
- Speak at a natural pace — not slower. Slow, halting speech reads as uncertainty. Match the speaker's cadence even if you're not 100% sure of every word.
- If you miss a word, KEEP GOING. Substituting a plausible word costs less than freezing or stopping. Partial credit beats no credit.
- Pronounce consonant clusters fully ('thr-ee', not 'tree'; 'asked', not 'ass'). Cluster simplification is the single biggest source of point losses.
- Lock in the sentence pattern in the first 1-2 seconds of your response. If you start uncertainly ('uhhh, I think she said…'), you've wasted your 10-second window.
- Don't whisper. The transcription model can't reliably parse low-volume speech. Speak clearly into the mic at a normal conversational volume.
Common mistakes
- Trying to translate the sentence into your native language while listening. By the time you finish translating, the audio has ended.
- Pausing for 2-3 seconds before starting. The recording timer is short — every second you hesitate costs you sentence-final words.
- Adding 'uhm' or filler words. Listen-and-Repeat is graded on accuracy — fillers count as errors.