Silence, Coverage, and the Wrong Kind of Politeness
Arjun, aiming for 115, records a mock speaking item: 30 seconds' preparation, then 90 seconds on 'Describe a skill you learned and explain why it mattered.' Analysis of the recording: 14 seconds of silence in the first 25; then a fluent, well-organised 50-second answer covering only how he learned to swim, never why it mattered; he stops at 64 seconds believing he has finished. In interactive listening he chose the longest, most formal option in 6 of 7 items, including 'Would you be so kind as to elaborate?' in reply to a classmate's casual question. Grammar accuracy in unplanned speech is 78%; in a rehearsed opening he uses, it is 96%. He asks whether he should memorise more openers.
1. Which of the two problems should be fixed first, and what makes the choice non-obvious?
Read the worked answer
Fix coverage first; fix register second - but the choice is genuinely contestable.
For coverage first: the prompt has two halves and he answers one. That is a relevance failure, not an accuracy failure, and it costs score on every extended-turn item he meets. It is also cheap to fix, because 26 seconds of unused speaking time is already available.
Against that: the register problem is systematic (6 of 7 items), and choosing the longest, most formal option every time suggests a single underlying rule - 'more elaborate is safer' - that will keep producing errors. One belief, many items.
What tips the balance: the silence. Fourteen seconds of dead air is the highest-cost, lowest-effort defect on the recording, and his 78% unplanned accuracy shows that the moment he speaks without a script, control drops. Attacking coverage forces him to speak unplanned, which also raises the accuracy issue. Register work is a separate lesson that can wait a week without compounding.
Caveat: if a single re-record shows coverage fixed but register unchanged, swap the order.
2. Rewrite his first 20 seconds. What should he be doing instead of staying silent?
Read the worked answer
Say the plan out loud; do not compose in silence.
A model opening: 'A skill I learned - swimming - and the reason it mattered is that it changed how I handle being bad at something. I'll say how I learned it, then what changed.'
That does three jobs in about eight seconds:
- It states the two halves of the prompt, which makes dropping the second half much harder later.
- It is a full, fluent, low-risk sentence, so the recording opens with language rather than a gap.
- It buys planning time inside the speaking turn, where it is scored, instead of before it, where it is not.
The general rule: use the preparation time for three keywords on the two halves, not for sentences. Silence in the first quarter of a response reads as a fluency problem even when the eventual answer is good, because a listener has nothing else to judge.
3. Should he keep the memorised opener? Argue both sides and commit.
Read the worked answer
For keeping it: it buys 96% accuracy in the first ten seconds, when raters form an impression; a confident opening reduces the chance that early disfluency colours the rest.
Against: it is only 10 seconds of a 90-second turn, so it cannot lift the overall sample much; it may delay real engagement with the actual prompt if the memorised line does not fit; and it gives him an excuse to avoid the unplanned speech that is his actual weakness.
Commit: keep one, use it once, and shrink it. Retain a short formulaic frame of six to eight words that can be adapted to any prompt ('The skill I'd pick is ___ , and it mattered because ___'). That keeps the accurate launch while forcing immediate content selection. Drop any opener longer than a sentence, and cap rehearsal at two minutes a day, because rehearsed fluency transfers poorly: his 78% unplanned figure is the number that predicts his score.