Replacing 'um' with silence, not with a better filler

30-second version: "Um," "uh," "like," "you know," "actually," "basically," "so" — these are what sayit's fluency scoring flags as filler words, and shaving up to 20% off your rate score when they're dense. The instinct to swap one filler for a "better" one (a strategic "well," a thoughtful "so") misses the actual fix: the gap they're filling can just be silent. A silent pause reads as composure; a filled one reads as searching, regardless of which word does the filling.

Filler words aren't a vocabulary problem — you're not missing a word, you're avoiding silence. "Um" isn't chosen for its meaning; it's a placeholder sound that fills the gap between "I need a second to think" and "I have my next word ready," because silence in that gap feels, to the speaker, like something has gone wrong. It hasn't. A brief pause while you retrieve a word is completely normal in fluent speech — the filler is an unnecessary addition to a pause that would have been fine on its own.

What actually gets flagged

sayit's fluency scoring detects a specific set of markers, case-insensitive and punctuation-stripped: um, uh, er, like, you know, literally, basically, actually, so. Filler density shaves up to 0.20 off the rate score — a gentle penalty by design, because occasional fillers are normal in real speech and heavy-handed punishment would push learners toward a different, worse habit: robotic, over-controlled speech that avoids all natural hesitation markers rather than just excessive ones.

Why swapping fillers doesn't fix anything

A learner who notices their "um" habit often "fixes" it by substituting a different word — starting every sentence with "so," padding every gap with "you know," using "actually" as a verbal tic instead of its actual meaning. This is worse than it sounds, because it feels like progress (the specific flagged word is gone) while the underlying habit — filling every thinking-gap with sound — is completely unchanged, and often harder to notice in yourself because the new filler doesn't trip the same self-awareness "um" used to.

The actual target isn't "stop saying um" — it's "let the gap after you'd normally say um be silent instead."

Why silence feels wrong and isn't

A pause of even half a second can feel, to the speaker, like an uncomfortably long silence — your own perception of time during a thinking-gap runs slower than a listener's does. To a listener, a clean half-second silence before you continue reads as composed, even thoughtful. The same half-second filled with "um" reads as visibly searching. The content of what you eventually say is identical either way — only the framing around the gap changes, and silence frames it better.

A practice routine

1. 1.Read a passage aloud and check your filler count and density in the fluency breakdown — a concrete number is more useful than a vague sense of "I say um a lot."
2. 2.Practice freeform speaking on a simple prompt with a deliberate silence rule: whenever you'd normally reach for a filler, close your mouth instead and let the half-second happen. It will feel unnaturally long the first several times. It isn't.
3. 3.Record the same prompt a few times across a week and watch the filler count trend down in Progress — a trending number is a better motivator than a single session's score.
4. 4.Don't over-correct into total pause elimination. The goal is a natural pause instead of a filled one, not the fastest possible delivery with no thinking room at all — see pauses: where fluent speakers breathe for where pauses genuinely belong.

Where this fits with pacing

Filler density and pace interact — rushing through a sentence you haven't fully planned is often what triggers a filler in the first place, because you've committed to speaking before you've finished thinking. Slowing very slightly and trusting a real pause is frequently more effective than "trying not to say um," because it removes the underlying rush that caused the habit rather than just suppressing its symptom.

Try it: record a minute of freeform speaking on any topic, check your filler count, then try the same topic again with a deliberate silence rule and compare.