Filler words: a practical list and how to use fewer of them

By the Dictalus team · Published July 12, 2026

Filler words are the sounds and phrases that hold your place while your brain catches up to your mouth. The short list: um, uh, er, like, you know, so (as a sentence starter), right, kind of, sort of, basically, actually. They are not a character flaw. They are a habit, and habits respond to reps. Below is the full list sorted by what each filler tells a listener, which ones cost you the most, and the one drill that cuts your rate.

The filler words list, by category

Not all fillers do the same damage. Group them by what they signal and each category points to its own fix.

"Like" floats across categories. It hedges ("it's like a database"), it hesitates ("it was, like, hard"), and it starts clauses. Count every use that isn't a literal comparison.

Which fillers matter most

Hesitation sounds read worst. Um and uh carry no meaning at all, so a listener hears them as pure signal: the speaker is searching. They also cluster at exactly the moments you can least afford them. Transitions between points. The first second after a hard question. Anywhere the stakes go up, the ums go up with them.

Hedges are quieter damage. "We should basically move the deadline" does not sound unprepared. It sounds uncommitted. A hedged recommendation invites a second opinion, and in a room full of opinions, the unhedged sentence usually wins. If you hear yourself stacking hedges ("I kind of think we should sort of consider it"), the problem is not delivery. You have not decided what you believe yet.

Approval-seekers erode authority a different way. Ending every second sentence with "right?" hands the listener a job: nod, or the talk stalls. Once or twice is rapport. As a tic, it reads as needing permission.

Here is the part most advice gets wrong: a few conversational fillers are normal human glue. Speech with zero fillers sounds rehearsed and slightly off, like a person reading. The target is lower density, not zero. Measure it as a rate, fillers per minute, and work on cutting your baseline in half. Expect the rate to spike under pressure. That spike is the thing you are training out.

Find your crutch word first

Most people do not use ten fillers. They lean on one or two, hard, and are usually wrong about which ones. The speaker who swears their problem is "um" often turns out to be a "so" starter with a "you know" habit.

So run the test. Record yourself for 60 seconds answering a plain prompt: explain what your team does to someone who just joined. No notes, no script. Listen back once and tally every filler. The tally does two things. It gives you a baseline rate you can actually track, and it tells you which specific word to listen for. You cannot fix what you do not hear, and once you know your crutch word, you will hear it everywhere.

The filler swap drill

The whole game comes down to one swap. When you feel a filler coming, close your mouth and pause instead.

There is a physical tell. Your mouth opens before the next thought has arrived, and the filler leaks out to hold the floor. The drill trains you to catch that half-second. Feel the um forming, close your mouth, let the silence sit, then continue when the thought is ready. The pause feels long to you. To the listener it reads as control. Nobody in the history of meetings has counted a speaker's pauses, but everyone counts the ums.

Run it as a rep: pick a prompt, speak for 90 seconds with one focus only, no filled pauses. Then record a second take back to back and beat your first count. Do not try to fix pace, structure, and fillers at once. One fix per rep. The swap stops being a decision after a couple of weeks of reps; it becomes what your mouth does.

Before and after

The same 10-second answer, first as most people say it, then with the swap applied.

Before: "So, um, I think the main takeaway is basically that we're on track, you know, and we just kind of need, uh, more time on the testing side, right?"

After: "The main takeaway: we're on track. [pause] Testing needs one more week."

Count the changes. The "so" wind-up is gone; the answer leads. "Basically" and "kind of" are gone, so the claim stands on its own. The "um" and "uh" became one deliberate pause. The trailing "right?" became a full stop. Same information, half the words, and the second version sounds like a person who has decided something.

That is the whole method. Know the list, find your crutch word, and drill the swap until the pause is the default. Fewer fillers is not about sounding polished. It is about removing the noise between your listener and your point.