Ten habits that hold strong English speakers a band below their level on TOEFL Speaking, why each is invisible from the inside, and the fix for each.
11 min readMonthly & Premium
Quick answer
Most capped Speaking bands come from habits rather than from weak English: stopping at 25 seconds, hedging instead of committing, memorised phrases, restating the question, monotone delivery, and treating Listen and Repeat as a warm-up when it carries 35 of the 55 raw points.
Why good English gets a middling score
The most frustrating message we get about Speaking goes something like this. My English is fine. I work in English, I watch films without subtitles, people understand me everywhere. My Speaking band came back lower than my Reading. What is going on.
What is going on is almost never a language problem. It is a set of habits that are invisible from the inside because they are not errors. Nothing in a hedged, twenty six second, evenly delivered answer is wrong. Every sentence would pass unremarked in conversation. It simply does not do the things the scale is built to detect, and because there is no error to find, the candidate reviews the recording, hears nothing wrong, and concludes the scoring is unfair.
This lesson is the list. Ten habits, each with why it is invisible, what it costs, and the smallest change that fixes it. Most candidates have three or four of them. Fixing two is usually worth half a band, and none of them requires learning any new English.
The ten habits
Habit
What it costs
The smallest fix
1. Stopping at 25 to 30 seconds
The largest single cause of a middling interview score, because a short answer is direct evidence of limited development.
One reason, taken down two levels of detail, ending with a specific occasion.
2. Hedging instead of committing
"It depends" costs five seconds and leaves the rest of the answer with nothing to develop.
Commit in the first sentence, qualify in the last ten seconds.
3. Treating Listen and Repeat as a warm-up
Ignores 35 of the 55 raw points in the section, the larger of the two blocks.
Ten minutes a day of listen once, repeat, check.
4. Memorised phrases and templates
Usually fits the question badly, and spends seconds of the window without adding content.
Memorise slots and timings, never sentences.
5. Restating the question
Burns five to eight seconds and adds nothing the rater did not already have.
A lead in that commits, then straight to the position.
6. Monotone delivery
No identifiable error, so it never gets noticed, and it drags every single item down together.
One strong beat per thought group, and let the pitch fall at the end.
7. Speaking quietly
Flattens intonation and blurs word endings on top of the audibility problem.
Speak at across-the-table conversation volume and check by listening back.
8. Self-correction loops
Two or three seconds per repair, plus a broken rhythm, for an error already recorded.
Repair inside the flow with "I mean", or let it go.
9. One idea said three ways
Reads as padding, which is scored as undeveloped rather than as length.
Replace the second restatement with a named example.
10. Never listening back
Guarantees that every habit above stays invisible indefinitely.
One transcribed answer a day, six minutes.
Ranked roughly by how much each one costs the average candidate.
The three that cost the most
Stopping early. Ask a candidate why their answer was twenty six seconds and the reply is always some version of: I said what I thought, and then there was nothing else that was true. That is a real problem and the solution is not more ideas, it is more depth on the idea you have. Any reason can go down two levels. What does it look like in practice, and when specifically did it happen. Those two questions turn twelve seconds of content into forty every time, and the detail they produce is precisely what the top of the scale asks for.
Hedging. "It depends on the situation" is usually the most accurate answer available and it is the wrong one to give. The scale rewards a developed position, not a fair description of your views. Commit early and add the nuance at the end, where it reads as range instead of as indecision. Note what this does not require: you do not have to believe your position more strongly, only to state it sooner.
Neglecting the repetition block. Seven items at five points each is 35 of the 55 raw points, and candidates routinely arrive at test day having practised interview answers for weeks and repetitions for approximately zero minutes. It is the more improvable block, it has a single correct answer per item so feedback is immediate, and it responds to ten minutes a day. If you are short of time, this is where the time goes.
Record one: the anti-hedge drill
Some people think that studying in a group is more effective than studying alone. Do you agree? Why or why not?
One rule for this recording: the words "it depends" are banned, and so is any sentence that presents both sides before you have taken one. In the first five seconds, say a lead in and then commit, plainly, using "I". Save every qualification for the final ten seconds. Listen back afterwards and time the exact second at which your position became clear. If it is later than second ten, run it again.
0:44
Model answer
The position lands at about second three. "I would say alone, for most of what I actually have to do" commits and scopes the claim in one sentence, with no preamble.
The mechanism is specific and slightly counterintuitive. Group study feeling productive without being productive is a real distinction, and it is content the prompt did not supply.
The concession is placed last and narrowed. "Where groups do work for me is right at the end" gives the answer nuance in the final ten seconds without ever putting the position in doubt. That is the same information a hedge would have carried, delivered where it helps instead of where it hurts.
Nothing is repeated. Each sentence adds something: the claim, the mechanism, the personal test, the outcome, the exception. Six sentences, six moves, about forty two seconds.
The habits nobody warns you about
Register drift. The interviewer asks about your weekend, and the answer comes back in the language of a written essay: "there are numerous factors which must be taken into consideration". This is not a higher level of English, it is the wrong level for the task, and a listener hears the mismatch immediately. Answer a conversational question conversationally. Plain first person speech reaches the top bands.
Practising only what you are good at. Almost everyone practises item 8 far more than item 11, because the personal recall question is pleasant and the abstract question is not. So the weakest slot stays weakest, and it stays weakest at the exact point in the block when concentration is lowest. Run the four questions as a block, always in order, always without stopping.
Chasing rare vocabulary. The instinct to raise a score by inserting impressive words is understandable and it backfires, because the words you control least well are the ones you are inserting, and misuse costs more than plainness does. Range means variety and precision, not rarity.
Reviewing by feel. "That one felt good" and "that one felt terrible" correlate weakly with the actual scores, because the feeling tracks how anxious you were rather than what the recording contains. Time the answer, count the silences, find the position sentence, count the concrete nouns. Numbers beat feelings here every time.
A one week habit audit
1
Day one: measure, change nothing
Record a full four question block. For each answer write down the length, the second at which the position appeared, the number of silences over two seconds, and your most repeated word with its count. Do not fix anything yet. You are building a baseline you can argue with later.
2
Day two: length only
Run the block again with one instruction: every answer reaches at least forty seconds through detail, not through padding. Ignore everything else, including quality. Length is the highest value habit and it is the easiest to change deliberately.
3
Day three: commitment only
Run the block again with hedging banned. Position by second ten in every answer. Again ignore everything else. Two habits changed in three days is a realistic pace, five is not.
4
Day four: the repetition block
Leave the interview alone entirely. Seven repetition items, listen once, repeat, check, and specifically monitor whether your last three words are as clear as your first three. This is the block carrying 35 of the 55 raw points and it is probably the one you have practised least.
5
Day five: transcribe one answer
Six minutes with a keyboard. Circle every vague word, every repeat, and every sentence that starts the same way as the one before it. This is the exercise that makes absences visible, which is the only reason these habits survive.
6
Day six: full section under test conditions
All eleven items, correct order, real setup, no stopping, no notes, after an hour of other work so your concentration is in the state it will actually be in.
7
Day seven: compare with day one
Same four numbers per answer. If length is up and silences are down, the band is moving whether or not any individual answer felt better. Trust the numbers over the feeling.
Record two: the anti-padding drill
Think about a time when a plan you made did not work out. What happened, and what did you do about it?
One rule: no sentence in this answer may restate a previous sentence. Every sentence has to add a new piece of information. That constraint will feel tight around second twenty five, which is exactly the moment when most candidates start padding. When you feel it, go to a smaller detail rather than to a bigger claim. Name a time, a place or a number.
0:44
Model answer
Every sentence is new information. The setup, the failure, the constraint, the cost, the response, the safeguard, the outcome. Seven distinct moves in about forty three seconds, with no restatement anywhere.
Both halves of the prompt are answered. What happened and what you did about it. The second half gets real weight rather than one throwaway clause, which is where two part questions usually lose points.
The detail is small and specific. April, four people, two days before, half the deposit, June, tickets paid that week. None of it is impressive and all of it is exactly the specific personal detail the top of the scale describes.
The close is honest rather than triumphant. "Though it was a smaller trip than the original one" ends on a real note in about three seconds. A manufactured upbeat conclusion would have taken the same three seconds and added nothing the answer had not already earned.
Spot the habit
Four recordings described. Identify what is actually costing the score.
A candidate speaks for the full 44 seconds with no pauses and no grammar errors, but the answer says the same thing three times in different words. What is the problem?
Development is about how far an idea is taken, not how many seconds were occupied. Three versions of one claim is one claim, and the middle of the scale describes exactly this: limited detail, however smoothly delivered.Filling the window is necessary and not sufficient. Padding a short idea out to 44 seconds protects the length and not the score.Nothing in the description suggests a vocabulary problem, and simple vocabulary used precisely scores well. The issue is that no new information arrives after the first ten seconds.No pace problem is described. The answer is smooth, which is part of why the candidate will not hear the issue when they listen back.
A candidate has practised interview answers daily for a month and has not practised Listen and Repeat at all. What is the likely effect on the section score?
Seven items at five points each is the larger share of the section. Leaving it untrained caps the band no matter how good the four interview answers become, and it is the block that responds fastest to practice.Harder to perform is not the same as worth more. The interview answers carry 20 raw points against 35 for the repetitions.Repetition is highly trainable, which is precisely why neglecting it is expensive. Chunking, phrasing and endings all improve within weeks of daily work.The skills overlap partly on delivery, but the memory and phrasing demands of repetition are specific and do not come free with interview practice.
Which self-review method is most likely to find these habits?
These habits are absences rather than errors, so they only become visible when you measure what should have been present. Four numbers per answer find them reliably in about two minutes.Hunting for mistakes is the method that guarantees you will miss all of these, because none of them is a mistake. This is why candidates review for months without improving.Model comparison helps, but without measurement you tend to notice vocabulary differences and miss the structural ones, which are the expensive ones.A friend will tell you whether your English is good, which you already know. They are unlikely to notice that your position arrived at second twenty two.
A candidate uses "in contemporary society, it is widely acknowledged that" to open every interview answer. What is the main cost?
The phrase is prepared, sounds different from the speech around it, consumes several seconds without content, and was written for a general essay rather than for a question about your own weekend. Delivery and development both take the hit.The vocabulary level is not the problem. The same words in a fitting, natural sentence would cost nothing.The phrase is perfectly grammatical, which is exactly why candidates trust it and keep using it.It is far too formal, not too informal, and formality itself is not penalised. What is penalised is prepared filler that carries no content.
The ten habit check, run it monthly
My answers reach 40 seconds or more through detail, not padding.
I commit to a position within the first ten seconds and qualify only at the end.
I have practised Listen and Repeat this week, not only interview answers.
Nothing I say sounds prepared or pasted in from a written source.
I never restate the question before answering it.
My voice rises and falls, and I have checked this by listening rather than by feel.
I speak at conversation volume, verified in a recording.
I do not stop to repair errors that would not confuse the listener.
Every sentence in an answer adds new information.
I have transcribed at least one of my own answers in the last week.
What to do with all of this
Do not try to fix ten habits. Pick the two that cost you the most, which for most candidates are stopping early and hedging, and work only on those for a fortnight. Habits compete for attention, and a candidate consciously monitoring six things at once produces stilted, careful speech that scores worse than the version with the original faults in it.
Then measure rather than judge. Length, position timing, silence count, most repeated word. Those four numbers per answer take two minutes and they will tell you honestly whether the fortnight worked, which no amount of listening back and feeling hopeful will.
And keep some perspective about the size of the thing you are fixing. Eight minutes, eleven items, 55 raw points. The habits in this lesson are worth real bands, and none of them requires you to become a better speaker of English than you already are. They require you to stop doing four or five specific things that felt like good practice and were not. That is a much smaller job than most candidates think they are facing, and it is why Speaking often moves faster than any other section once you start working on the right thing.
Monthly & Premium
9 more sections in this lesson
You have read the opening. The rest covers the method in full, with worked examples and practice questions that explain why each wrong answer is wrong.
All 45 lessons: Reading, Listening, Writing and Speaking
Real exam audio and record-yourself speaking drills
Usually because of habits rather than language. Answers that stop at 25 seconds, hedge instead of committing, or repeat one idea in three ways contain no errors at all, which is why reviewing your own recordings for mistakes never finds the problem. Measure length, position timing and silences instead.
Stopping early. An answer of 25 to 30 seconds against a 44 second window is the clearest possible evidence of limited development, and no amount of accuracy compensates. The fix is depth on one reason rather than a search for more reasons.
Memorised sentences are. A memorised structure is not. Learn the slots and their timing (lead in, position, mechanism, specific detail, close) and fill them fresh each time. The problem with prepared sentences is not that a rater spots them, it is that they answer the question you rehearsed rather than the one you were asked, and they use up a window that is only forty four seconds long.
In a 44 second window, yes. It is often the most honest answer and it costs you several seconds while leaving the rest of the response with nothing to develop. Commit in the first sentence and put the qualification in the last ten seconds, where it reads as range.
More than you currently do. Seven repetition items carry 35 of the 55 raw points, against 20 for the four interview answers, and most candidates arrive having practised the interview almost exclusively. Ten minutes a day on repetitions is the highest return time in the section.
You cannot tell while speaking, which is why it survives. Record an answer, then hum it back with no words. If the hum is close to a flat line, the delivery is flat, and the fix is one clear strong beat per thought group with the pitch falling at the end of each sentence.