How should I use pauses when speaking?

How much silence does good speech have?

Between 18% and 35% of total speaking time — that is the pause-ratio band the strongest measured delivery sits in. Barack Obama’s measured address is 21% silence: 58 deliberate stops in 193.6 seconds, roughly one every three seconds. A fifth of the address is nothing at all — on purpose.

Obama is not the outlier — he is the pattern. The heaviest deliberate pausers we’ve measured all cluster in the same band, topped by John F. Kennedy, whose “ask not” passage reads 28.8% silence — the highest pause ratio in our library, and nearly a third of the most quoted stretch of American oratory. The pattern is not a law: real counterexamples exist — Walter Cronkite commands at 1.4% silence, Adele at 5.9% — but they command on depth and pace. The pause is the lever that costs nothing.

Why is silence powerful in a speech?

A pause ratio under 8% is measurable rushing: the words arrive faster than a listener can process them, and every gap gets filled before it can do any work. A deliberate stop does the opposite — it gives the room time to absorb the point you just made, and it sounds like a speaker who is not afraid of the quiet.

Rushing shows up as a pair of readings — a fast speaking rate with a starved pause ratio — and the fix for one is usually the fix for the other. What the library shows is that the voices people describe as commanding and unhurried are, measurably, the ones that stop most: silence is the cheapest authority signal there is, because it costs nothing and rushing is what everyone does instead.

How long should a pause be?

0.4 to 1.0 seconds is the optimal band for average pause length. The measured speakers who pause best hold roughly two-thirds to three-quarters of a second: Barack Obama averages 0.71 s, John F. Kennedy 0.78 s, Maya Angelou 0.64 s. Much past a second and a half, a pause stops reading as emphasis and starts reading as losing your place.

The band is wider than most speakers dare use. Alan Rickman — 19% silence overall — holds one measured pause past 2.3seconds and never fills it; Obama’s longest stop is a brisk 1.33s. Both work. What doesn’t is the micro-gap: a pause under a quarter of a second is a breath, not a beat, and listeners never register it as a choice.

Do pauses replace filler words?

Yes — under 2 fillers a minute goes unnoticed, above about 4 a minute it distracts, and most “um”s are simply a gap the speaker was not willing to leave silent. A silent beat does the same job as an “um” — holding your turn while you find the next thought — but reads as confidence instead of hesitation.

This is why we treat pausing and fillers as one skill, not two. Trying to suppress “um” directly just makes you self-conscious; putting a real pause where the filler wanted to go removes the need for it. The pause budget is generous — up to a third of your total time — so there is always room to trade a filler for silence.

How do I practise pausing?

One drill, about a minute: read the same 20-second passage aloud twice — once at your natural pace, once stopping completely at every full stop for a silent “one-and” — and measure both takes. The gap between the two pause ratios is usually the whole lesson.

The first take tells you your habit; the second tells you your range. Most people’s natural read lands under 18% and feels normal from the inside — which is exactly why measuring beats guessing. Aim the deliberate take at the 18–35% band with stops near the 0.4–1.0 s mark, then listen back to both: the slow version almost never sounds slow from the outside. The full band table, and how every number on this page is computed, is on the methodology page.

What the instrument can’t hear

We measure how much you pause and how long — not where. Christopher Walken’s pause numbers are textbook; the famous strangeness is entirely in the placement, which no acoustic metric scores. Our library figures also come from short public recordings, and pause detection needs clean silence: on old, noisy transfers a speaker’s real pauses can vanish into the crackle and read as zero. And a microphone cannot tell a pause held for effect from a speaker who lost the thread — only the room can. Treat these numbers as the mechanics of good pausing, not the whole art.