Skip to content

Pace, pauses, filler words & clarity

At a glance

Priority: P0 · Complexity: ⅗ · Est. time: 2 h · Phase: 1 · Prereqs: Pronunciation: word stress, rhythm & intonation You're done when: a 3-minute recording of you explaining a design shows under 4 fillers per minute, a deliberate pause at every sentence boundary that matters, and a pace between roughly 130 and 160 words per minute, measured from a transcript.

Drilled in Week 1 (pace and pausing), Week 4 (cutting fillers) and Week 11 (calls and video).

Why it matters

Fluency problems in a fluent, advanced speaker are rarely about vocabulary. They are about planning under time pressure: the brain starts a sentence before it knows the end, so it fills the gap with "um", "so", "basically", or it speeds up to escape the silence. Listeners then hear low confidence, even when the content is excellent.

Three facts that reframe the problem:

  • Pauses are not failures. Skilled speakers pause often; the pause is where the listener processes what you just said. Silence feels several times longer to the speaker than to the listener.
  • Fillers are not the enemy; unplanned ones are. Occasional fillers are normal in every language. The problem is density (more than about 5 per minute) and placement (in the middle of a phrase, rather than between ideas).
  • Speed is not clarity. For non-native listeners on a compressed call, a rate above about 170 wpm in technical content reduces comprehension. Native rates in conversation often reach 150-190 wpm, but presenters usually aim lower.

Staff-level speaking is mostly structure plus pause: know the first sentence, know your three points, stop between them. See Structuring spoken answers.

Core concepts

1. Pace: numbers to calibrate against

Context Target words per minute (wpm) Notes
Casual conversation 150-190 Native speakers; do not aim for this on calls
Presentation / design review 120-150 Enough time to show slides and let numbers land
Explaining something complex, to a mixed audience 110-130 Slow for the key term or number
Audiobook narration 150-160 Steady benchmark
Poor audio or many non-native listeners 100-120 Plus pauses and chat summary

These are commonly quoted ranges, not laws; measure yourself. To compute: transcribe 1 minute (any auto-captions), count words. Most people are surprised: they feel slow at 165.

Pace is variable, not constant. Speed up for context and connectors; slow down for numbers, names, decisions and negatives. "The error rate went from two percent to eleven percent, in four hours." The bold portion is where you slow.

2. The pause toolkit

Pause Length Use
Comma pause ~0.3 s Between clauses in a long sentence
Sentence pause ~0.7-1 s After every full sentence in a technical explanation
Section pause 1-2 s Before a new topic; after a key claim
Power pause 2-3 s After a question to the audience or before a decision ask

Techniques:

  • Breathe on the pause. Breath resets rhythm and stops the run-on. Silent inhale, then start.
  • Land the sentence with a fall. A falling intonation (see the pronunciation page) makes the pause feel intentional; a rising or level pitch makes it feel like hesitation.
  • The Sentence Stop. Practise ending sentences at a real full stop rather than adding "and" plus another clause. Replace "and... so... which..." chains with two sentences.
  • Silent planning. Use the pause to decide the next sentence. A visible thinking pause beats a filler.

3. Fillers: types and replacements

Type Examples Why it happens Replace with
Hesitation sounds um, uh, er processing gap silent pause (close the mouth)
Discourse padding you know, I mean, kind of, sort of seeking agreement / softening drop it, or a precise hedge ("roughly", "in most cases")
Starter filler so, well, basically, actually, right starting before planning a stock opener ("Here's the short version:")
Trailing filler ... or something, and stuff, like, right? not finishing the thought finish with a clear last word, then stop
Repetition "we need to, we need to, we need to..." restart loop pause, then say the whole clause once
Verbal tics of intensity literally, honestly, obviously emphasis habit remove; obviously can sound condescending

Note "like" as a filler is common in American informal speech; in a senior, formal setting it reads as junior. "Basically" is often a signal that you are about to simplify: do it, but no need to say it.

4. Cutting fillers: the process

  1. Measure. Record 3 minutes, get a transcript, count fillers. Compute fillers per minute. (Auto-caption tools often delete fillers, so play the audio and tally by ear.) Tools: Otter, Zoom/Teams transcripts, or a phone tally.
  2. Identify your top two. Most people have 1-2 dominant fillers. Attack only those.
  3. Replace, not suppress. Suppression increases anxiety. Replace with a pause plus breath.
  4. Slow down 10%. Fillers spike when speed exceeds planning speed.
  5. Plan the first sentence. Most fillers sit at the start of a turn. Prepare openers (see phrase bank below).
  6. Practise with constrained tasks. 60-second talks on a random work topic; the buzzer rule is "restart at each filler" (Toastmasters "ah counter").

5. Clarity: the sentence-level habits

  • One idea per sentence when speaking. Written sentences of 25+ words are hard to say and hard to follow.
  • Lead with the point. "We should defer the migration because two dependencies aren't ready." not "So there are a number of factors here, and given that we've looked at..."
  • Signpost. "Three things. First... Second... Third..." Listeners cannot skim; counting helps memory.
  • Repeat the key number or decision in different words once.
  • Avoid nested asides. "The service, which was built by the payments team three years ago before the reorg, that we ..." becomes two sentences.
  • Prefer concrete words. "Latency went from 800 ms to 200 ms" over "significant improvement".
  • Check understanding: "Does that match your understanding?" is better than "Does that make sense?", which can sound patronising.

6. Phrase bank: openers, transitions, buying time

Function Phrases
Open an answer "The short answer is..." · "Two things stand out." · "Let me answer in three parts." · "The trade-off here is..."
Buy time (without "um") "That's a good question; give me a moment to think." · "Let me think about how to frame this." · "There are a couple of ways to look at it." · "Let me come back to that in a second."
Restart cleanly "Let me rephrase that." · "Let me put that another way." · "What I mean is..."
Signpost "First... Second... Finally." · "Turning to..." · "To summarise..." · "Before I go on, ..."
Slow down for a key point "The important number is..." · "What matters most is..." · "Here's the decision I need."
Check "Does that answer your question?" · "Should I go deeper on any of that?"
Finish "That's my recommendation." · "So, to sum up, ..." · "Happy to take questions."

7. Call and video-call clarity tactics

  • Latency and overlap. Video adds 200-500 ms of delay, so wait about a second after someone stops before you start; your instinct to jump in is more likely to cause overlap.
  • Mute discipline. Mute when not speaking; unmute before you begin and take a breath so you do not clip the first word.
  • Front-load and summarise in chat. After you speak, paste a two-line summary or the decision in chat.
  • Numbers and names. Spell out or type: "Kafka, K-A-F-K-A", "seven zero, that's seventy percent".
  • Camera and mic. Mic 10-15 cm from mouth; look at the camera when making the key point.
  • Repair. If you are told "you cut out", do not repeat everything; give the headline again in one sentence.

Resources

Resource Type Why this one Level Cost
Julian Treasure, "How to speak so that people want to listen" (TED) video Pace, pitch, pause and register in 10 minutes; good for a first self-audit intermediate free
Toastmasters International community The "ah counter" role is a proven fillers feedback loop; clubs run online all freemium
BBC Learning English audio/video 6 Minute English gives a calibration of natural conversational pace and filler use in the UK intermediate free
Otter tool Transcribes meetings; read your transcript to count fillers and words per minute all freemium
YouGlish interactive Search "let me think" or "the short answer is" to hear real pauses and stress all free
Rachel's English video Rhythm and the role of pauses in American English intermediate free/freemium
TED 💎 video Pick 3 engineering talks and count wpm: the best ones run around 130-160 wpm with visible pauses intermediate free
Zoom or Teams cloud transcript tool Free wpm and filler data from your own calls, and the best real-world evidence all free

Hands-on lab (60 min)

  1. Baseline (15 min). Record 3 minutes: "Explain your last production incident to a new VP." Do not prepare.
  2. Measure (10 min). Transcribe (auto), count words and divide by minutes; tally fillers by ear. Note your top two fillers and where in the sentence they occur (start, middle, end).
  3. Design your openers (10 min). Choose 3 opener phrases from the phrase bank and write your own for "what happened / why / what next".
  4. Drill with constraints (15 min). 4 x 60-second talks: same topic; (a) 130 wpm, (b) a pause after every sentence, © the buzzer rule: restart at any filler, (d) 3 points with signposts.
  5. Re-record (10 min). Same prompt as step 1. Compare pace, filler density, pause count.

Expected output: filler rate down by at least 40% and an explicit list of your two fillers with replacements.

Questions

L1 — Recall

Q1. What is a reasonable pace range for a design review, and how should pace vary within a talk?
Answer

About 120-150 wpm; slower (100-120) with mixed or non-native audiences or poor audio. Slow down on numbers, names, decisions and negatives; speed up slightly for context and connectors.

Q2. Name four categories of filler.
Answer

Hesitation sounds (um/uh), discourse padding (you know, kind of), starter fillers (so, well, basically), trailing fillers (or something, and stuff), repetition/restarts, intensity tics (literally, honestly). Any four.

Q3. What should replace a filler?
Answer

A silent pause with a breath (or a precise buy-time phrase). Suppressing without replacing raises anxiety and the filler returns.

Q4. Why is "Does that make sense?" sometimes a poor check question?
Answer

It can sound condescending, implying it might not, and invites a polite "yes". Better: "Does that match your understanding?" or "What questions do you have?".

L2 — Apply

Q5. Rewrite as two clean spoken sentences with a fall: "So basically what we're kind of seeing is that the latency, you know, spikes when the cache, um, gets cold and then the database is like overloaded."
Answer

"Latency spikes when the cache goes cold. The database then gets overloaded." Each sentence has one idea and ends on a fall; all fillers are removed and "kind of" becomes unnecessary because the claim is factual. If you are not sure, hedge precisely: "Latency appears to spike when..."

Q6. You are asked a hard question and need time. Give three ways to buy 3-5 seconds without "um".
Answer

"That's a good question; let me think about how to frame it." · "There are a couple of ways to look at this. Let me start with the trade-off." · Repeat the question: "So you're asking whether we can hit the deadline without the migration?" (also confirms understanding).

Q7. Your transcript shows 210 words in 90 seconds. What is your pace, and what do you do?
Answer

210 / 1.5 = 140 wpm: in range for a presentation. If the audience is non-native, go to ~120 and add pauses. Pace alone is not the problem; check whether pauses are present.

Q8. Write the opener and the first two signposts for "why we should adopt event sourcing for the shipment tracker" in under 40 words.
Answer

"Short answer: yes, but only for tracking. Three reasons. First, auditability: every status change is kept. Second, replay: we can rebuild views. Third, cost: it adds complexity, so I'd limit it to one service." Recommendation first, numbered points, a caveat.

L3 — Judge and choose

Q9. A peer says fillers are "natural" and you should not worry. Another says you must eliminate them. What is the evidence-based position?
Answer

Both are partly right. Fillers appear in all fluent speech; the aim is density and placement, not zero. Reduce them where they cost credibility (openings of answers, in the middle of key statements, more than ~5/min). Speakers who pause deliberately are rated as more confident and clear than speakers who eliminate pauses.

Q10. You speed up when nervous. Two fixes: (a) tell yourself to slow down, (b) build in structural pauses. Which is better and why?
Answer

(b): willpower fails under stress. A plan that pauses at each signpost or section boundary works without conscious control. Combine with a planned first sentence and a written 3-point skeleton.

Q11. Is it OK to say "like" and "you know" in a stand-up with your peers but not with executives? Defend.
Answer

Register varies: peers tolerate more; executives read the same tics as low confidence. The safer habit is to train one register: the clear one. Otherwise your worst-case habit appears in your highest-stakes moment.

L4 — Staff-level

Q12. In an incident bridge a VP interrupts your update, and the call has 30 people, poor audio and multiple non-native speakers. How do you communicate?
Answer

Stay slow, factual, structured. "Status: mitigated, not resolved." Pause. "Impact: shipments in EMEA delayed by about 40 minutes." Pause. "Next update in 15 minutes." Post the same three lines in chat. If interrupted, finish the sentence, then answer. Do not speed up to defend yourself; pace is a signal of control.

Q13. A junior engineer's fillers make them seem unsure, and the team lead wants you to "fix" them. How do you coach?
Answer

Ask permission, then use data: record a 2-minute demo together and tally fillers. Choose one target (their top filler). Give the replacement (pause + breath) and a structure (point, reason, example). Practise 60-second updates weekly. Praise progress with numbers. Do not correct in public. See Feedback, disagreement & conflict.

Q14. You are presenting to the board in 5 minutes and feel nervous. What two things do you do in the first 20 seconds?
Answer

Deliver a memorised first sentence (your headline) and take a full breath before it and after it. The memorised opening removes the highest-filler moment; the pause sets the pace for the entire talk.

Real-world use cases

  • Incident bridge updates: a 3-line status with pauses; repeated in chat.
  • Design review: signposted trade-offs so reviewers can interrupt at boundaries, not mid-thought.
  • Stakeholder call across time zones: slower pace and a written recap for people using captions.
  • Interview: deliberate pause before answers ("let me think for a moment") reads as senior, not slow.
  • Demo: narrating while clicking: fillers spike; pre-script the first sentence of each step. See Meetings, presentations & speaking on calls.

Pitfalls & anti-patterns

  • Trying to eliminate all fillers at once instead of the top two.
  • Replacing a filler with a new tic ("right?", "obviously").
  • Racing to fill silence after a question; the silence is yours.
  • Recording without listening back; you cannot fix what you do not hear.
  • Slowing all speech evenly, producing a flat, slow monotone. Vary pace.
  • Using auto-captions that delete fillers as your measure.
  • Apologising for pauses ("sorry, just thinking").

Checklist

  • I know my wpm and my top two fillers, with numbers.
  • I have 3 opener phrases memorised and used them on live calls.
  • I pause and breathe after every key sentence.
  • I slow down for numbers and negatives, and repeat decisions once.
  • I put a two-line summary in chat after a key contribution.
  • I answered all L3 questions out loud in < 3 min each.