Speaking
Listen to a sentence and repeat it exactly as you heard it.
Listen, then repeat the sentence exactly.
The seminar has been moved to the lecture theatre on the second floor.
Twelve words is within normal working memory when held as one idea. Repeating the stress pattern recalls the words themselves.
The audio plays once, so give it your full attention rather than starting to rehearse partway through. Hold the sentence as one idea, since a meaningful clause survives in memory far longer than a list of unconnected words does.
As you listen, register where the speaker stressed and paused. That contour is a retrieval cue: replaying the tune of the sentence in your head usually brings back the words attached to it, including the ones you did not consciously catch.
Start as soon as the recording ends, while the sentence is still fresh. Every second you spend assembling it silently loses words and eats into a response window of only 15 seconds, which is barely longer than the sentence itself.
If you only caught a fragment, speak that fragment fluently and confidently rather than stopping. Content is scored on the words you produce, and this task also feeds your listening score, so silence gives away marks in two sections at once.
Listen for meaning, not individual words — the sentence is far easier to hold in memory as a single idea than as a string. Mirror the stress and intonation you heard; matching the rhythm helps recall the words themselves.
Staying silent because you did not catch everything. Partial answers earn partial content marks, and saying nothing scores zero on all three traits.