What you'll see
Get familiar with the task.
You will hear a short academic recording ONCE. A transcription is shown on-screen with several words replaced by similar-sounding or similar-looking wrong words. Click ONLY the words that differ from what the speaker says. Audio starts after a 10-second countdown — no pause, no replay. Scoring is +1 for each correct click, −1 for each incorrect click (minimum 0 per item) — so do NOT guess.
You will hear a recording. Below is a transcription of the recording. Some words in the transcription differ from what the speaker(s) said. Please click on the words that are different.
You will hear a recording. Below is a transcription of the recording. Some words in the transcription differ from what the speaker(s) said. Please click on the words that are different.
Press Start to begin
3 strategies that work
The mindset before the click.
- 1Strategy 1
SOUND-OUT content words in the 10-second window
Skim only the NOUNS, VERBS, ADJECTIVES and ADVERBS. Silently sound them out — knowing the pronunciation makes the mismatch obvious when you hear the audio.
- 2Strategy 2
LISTEN for every content word — even the de-emphasised ones
Wrong words are often de-stressed or said quickly. Focus your ear on every content word, not just the obvious stressed ones.
- 3Strategy 3
READ along + click ONCE — do NOT re-read after the audio
Eyes track the text as audio plays. Click the moment you hear a mismatch. After audio ends, STOP — don't second-guess with grammar/logic. Wrong clicks lose a point.
Try each strategy
Guided practice you can do alone or with a partner.
1SOUND-OUT content words in the 10-second window
10 seconds isn't enough to read the whole passage — and you don't need to. The wrong words are almost always CONTENT words (nouns, verbs, adjectives, adverbs) chosen because they look or sound similar to the real word: equations / reactions, arrowhead / arrow, reality / nature. If you've already pronounced each content word in your head, your ear catches the swap instantly.
- Skim line-by-line. Ignore THE, A, AN, OF, IN, ON, AT — they're never wrong.
- Pronounce each content word silently. If you stumble on a word, mark it mentally — that's exactly where the trap is.
- Watch for word-family pairs you already know are commonly confused: affect/effect, allusion/illusion, climactic/climatic, credible/credulous, ingenious/ingenuous, official/officious, reluctant/reticent.
- Pair drill: dictate a list of 10 'sound-alike' pairs (accept/except, jealous/zealous, south/north, colonize/conquer). Partner writes them down — check spelling AND meaning.
- Pair drill: read Q1 silently for 10 seconds. Then close eyes — recall as many content words as you can. Builds the 'sound + image' link.
2LISTEN for every content word — even the de-emphasised ones
Examiners don't put wrong words on the stressed syllables — too easy. The traps are on words that the speaker says fast, quietly, or in the middle of a clause. You must train your ear to weigh every content word equally, not just the loud ones.
- Don't drift between sentences — wrong words can land anywhere, including the very first or very last word.
- If you miss a word, KEEP MOVING. Don't backtrack mentally — you'll miss the next 3 words too.
- Stressed content words = topic markers; de-stressed content words = traps. Listen to both.
- Pair drill: read aloud a passage with 3 hidden swaps. Partner listens once and points to the wrong words.
- Pair drill: take a 60-second clip. Partner reads it twice — first naturally, second deliberately de-emphasising 2 content words. You spot which were swapped.
3READ along + click ONCE — do NOT re-read after the audio
This task scores +1 for each correct click and −1 for each wrong click (minimum 0). Over-clicking is the single biggest scoring leak. The rule: click only when you HEARD a mismatch, in real time. After the audio ends, DO NOT scan the passage to 'fix things' using grammar or logic — wrong words usually fit grammatically and logically (that's why they were chosen).
- Sync your eyes to the audio — track the word being said. The instant you hear a mismatch, click.
- If unsure between two candidates, click NEITHER. Half-confidence = expected score of −0.5.
- Exception: if you spot a clear factual mismatch (e.g. a proper noun mentioned earlier reappears with a different spelling later), correcting it is safe.
- Aim for the EXPECTED number of swaps — usually 5–8 in a 70–90 word passage. If you've clicked 10+, you're over-clicking.
- Pair drill: do Q1, write down ONLY the words you'd click. Partner tells you which clicks were correct. Calculate your net score (right − wrong) to feel the cost of over-clicking.
Press Start. Use the 10-second countdown to sound-out content words, listen ONCE, then click every word that DOESN'T match the audio. Don't re-read after the audio ends. Reveal the answer key after you finish.
Reveal correct mismatches + reasoning
8 wrong words in the transcription. The speaker actually says: 1. 'chemical equations' → 'chemical REACTIONS' (the talk is about reactions, not equations). 2. 'our arrowhead' → 'our ARROW' (one syllable difference). 3. 'occur in reality' → 'occur in NATURE' (similar idea, different word). 4. 'arrow facing from right to top' → 'arrow facing from LEFT TO RIGHT' (3 swaps in a row — the speaker repeats his earlier phrase 'from left to right'). 5. 'reacting to form produce' → 'reacting to form PRODUCTS' (singular/plural shift — easy to miss because 'produce' sounds close). 6. 'a state of equality' → 'a state of EQUILIBRIUM' (the actual chemistry term). Why the traps work: every wrong word is grammatically valid and topically plausible. Only your ear (not your logic) can catch them — which is exactly why Strategy 3 says don't re-read after the audio. Scoring: +1 per correct click, −1 per incorrect click (minimum 0 per item). Max for Q1 = 8 points.
Do it under time
Simulate the real test.
Same exam rules: 10-second countdown, audio plays ONCE, click ONLY the words that differ from the audio. Wrong clicks lose a point — when in doubt, leave it alone.
Reveal correct mismatches + reasoning (Q2)
7 wrong words in the transcription. The speaker actually says: 1. 'fundamental rift' → 'fundamental SHIFT' (rift / shift — one consonant apart). 2. 'manufactured everything' → 'manufactured SOMETHING' (sound similar; both fit the sentence). 3. 'really effective' → 'really EFFICIENT' (the classic effective / efficient confusion). 4. 'there was centralization' → 'there was STANDARDIZATION' (both -ation nouns, both plausible in an industrial-era context). 5. 'a set of presumptions' → 'a set of ASSUMPTIONS' (presumption / assumption — easy swap). 6. 'things are functionally unstable' → 'things are FUNDAMENTALLY unstable' (both adverbs, almost the same shape). 7. 'economies of recovery' → 'economies of DISCOVERY' (recovery / discovery — rhyming pair). Every trap word fits grammatically and is topically related — proof that you can only catch them by listening, not by reasoning. Strategy 3 again: do not re-read after the audio ends. Scoring: +1 per correct click, −1 per incorrect click (minimum 0 per item). Max for Q2 = 7 points.
You will hear a recording. Below is a transcription of the recording. Some words in the transcription differ from what the speaker(s) said. Please click on the words that are different.
Press Start to begin
Tick each box only when you can do it without looking back. This unlocks your next chapter.