Why Filler Words Happen and How to Cut Them in Half

You say "um" about 2-3 times per minute in normal conversation. You probably think it's closer to once. That gap between perception and reality is where the entire filler word problem lives.

Here's the thing that changes how you think about this: filler words aren't a bad habit. They're a feature of human speech production. Every language on Earth has them. And most advice about eliminating them is wrong. Understanding how fillers work is one of the key dimensions of effective communication.

What are filler words, actually?

Linguists split them into two categories that work very differently in your brain.

Filled pauses -- "um" and "uh" -- are sounds your brain inserts where silence would otherwise be. They exist in every known human language. Japanese has "eto" and "ano." French has "euh." Mandarin has "nage." Your brain produces them automatically, not by choice.

Discourse markers -- "like," "you know," "so," "basically," "I mean," "right" -- do something different. They manage the flow of conversation: signaling relationships between ideas, holding your turn, indicating you're about to rephrase.

They're more cultural and generational than filled pauses.

Both get lumped together as "disfluencies." That label is misleading. Thirty years of linguistics research has consistently shown these aren't errors in the system. They're part of how the system works.

Why does your brain produce them?

The dominant model in psycholinguistics, built by Willem Levelt and refined by dozens of researchers since, describes speech as a four-stage pipeline running in parallel:

Conceptualization -- forming the idea
Formulation -- selecting words, building grammar, planning sounds
Articulation -- physically producing the speech
Self-monitoring -- checking your output against your intention

This pipeline processes two to three words per second, each pulled from a vocabulary of tens of thousands. About 150 words per minute in normal conversation. A filler word appears whenever there's a bottleneck anywhere in the pipeline.

Your word-finding system is lagging

The most common trigger for "um" and "uh" is a delay in lexical retrieval. Your brain has the concept but hasn't found the word yet.

Herbert Clark and Jean Fox Tree at Stanford published a landmark study showing that "um" and "uh" aren't random noise -- they're signals with distinct meanings. "Uh" tends to precede shorter delays (the word is almost ready). "Um" signals a longer delay (the word needs more time). Listeners actually use these signals. When you say "um," your listener unconsciously adjusts their expectations and waits.

In this sense, fillers are cooperative. They tell your listener: "I'm still talking, still thinking, don't jump in yet."

Higher cognitive load means more fillers

This is one of the most reliable findings in speech research. Filler rate tracks cognitive load almost linearly:

Simple, rehearsed speech: almost no fillers
Familiar topic, spontaneous speech: moderate fillers
Unfamiliar topic, spontaneous speech: more fillers
Speaking while doing mental math: the most fillers

This is why you use more fillers in a live Q&A than a rehearsed presentation. More in your second language than your first. More when you're tired than when you're fresh.

It's a resource allocation problem, not a discipline problem -- closely related to the challenge of thinking faster when speaking.

Some fillers are actually doing work

Most articles will tell you every "like" is verbal garbage. That's wrong.

"You know" often invites shared understanding: "The meeting was, you know, one of those where nothing gets decided." It asks the listener to fill in from shared experience.
"Like" functions as a quotative ("She was like, 'absolutely not'") or a hedge softening an approximation ("It took like three hours").
"I mean" signals you're about to rephrase or clarify.
"So" marks a transition from setup to conclusion.

Not every instance is doing useful work. But painting them all as empty tics misses the reality of how conversation actually functions.

When do fillers actually hurt you?

This is where most filler word advice goes sideways. "Eliminate all filler words" is the standard prescription, as if they're universally toxic. They're not. Context determines everything.

Where fillers cost you

Formal presentations and speeches. When you're the sole speaker and the audience is evaluating your credibility, frequent fillers measurably hurt you. Studies in communication journals have found that speakers with higher filler rates get rated as less credible, less prepared, and less knowledgeable -- even when the content is identical to a low-filler version. Same words, different perception.

Job interviews and evaluative conversations. Fillers create a gap between what you know and how competently you appear to know it. Two candidates with equal qualifications -- the one who pauses cleanly between thoughts sounds more confident than the one who bridges every gap with "um, so, basically."

Recorded content. Podcasts, videos, voice messages. Fillers that feel natural in real-time become friction when someone is choosing to spend their limited attention on your content.

When they cluster. Isolated fillers are invisible. Stacked fillers -- "So, um, basically, like, what I was, you know, trying to say..." -- create a sense of incoherence even when the underlying thought is perfectly clear.

Where fillers are fine

Casual conversation. In everyday dialogue, moderate filler use is normal and often functional. Trying to eliminate all fillers from casual speech makes you sound like a malfunctioning robot. Nobody wants that.

Turn-holding. In group conversations, a quick "um" or "so" signals you're not done speaking. Without it, someone jumps in and you lose the floor.

Softening. Sometimes "like" does real diplomatic work. "That idea is like, not quite right" lands softer than "That idea is not right." In contexts where social harmony matters, that softening is intentional, not sloppy.

The practical goal: fewer fillers where they hurt. Don't waste energy policing them where they're harmless.

How to actually cut filler words in half

The research points to specific, evidence-based techniques. The most effective ones share a thread: they work with your speech production system, not against it.

Step 1: Build real awareness

You can't fix what you don't notice. And you almost certainly don't notice half your fillers.

In controlled studies where speakers estimated their filler rate before listening to recordings, they typically guessed 30-50% of their actual count. The gap is enormous.

Record yourself in a real conversation. Not a rehearsed speech -- an actual, unscripted conversation or explanation. Play it back and count. Most people describe this experience as "eye-opening" or something less printable.

Here's what's interesting: awareness alone moves the needle. Research on self-monitoring in speech production shows that once speakers become conscious of their filler patterns, the rate begins dropping even without deliberate effort.

The self-monitoring system, which normally runs in the background, starts catching fillers before they reach your mouth.

Studies have shown awareness-based training cuts filler usage by roughly half within a few weeks of consistent practice. Consistent is the key word -- one recording session creates temporary awareness. Regular practice creates lasting change.

Step 2: Replace, don't suppress

Most people try the willpower approach: "I will not say um." This backfires predictably.

Suppression adds cognitive load. You're now monitoring for fillers on top of everything else your brain does while speaking. That extra load can paradoxically increase filler usage.

You're fighting the system instead of working with it.

The effective approach is replacement. Instead of trying not to say "um," practice replacing the filler with silence.

The process:

You feel the impulse to fill a gap with "um"
You say nothing. Just silence. Even half a second.
The next word arrives (it almost always does within a beat)
You continue

The counter-intuitive part: the silence feels excruciating to you. To your listener, it sounds confident. What registers as a two-second void in your head registers as a measured, deliberate pause to everyone else. Professional speakers, news anchors, and trial lawyers use silence constantly. It's one of the most powerful tools in communication, and most people are afraid of it.

Step 3: Lower the cognitive load

Since filler rate tracks cognitive load, reducing the load reduces the fillers. The same strategies that prevent blanking mid-sentence (covered in our companion article) apply directly.

Chunk your thoughts. Speak in short, complete ideas instead of long, winding sentences -- this is also the core technique for stopping yourself from rambling. Each chunk gives your brain a natural pause point to plan the next thought -- no filler needed to hold your place.

Use frameworks. "There are three things I want to cover. First... Second... Third..." A roadmap means your brain knows what comes next in the structure. When you know what's next, you don't need a filler to buy planning time.

Slow down 10-15%. Most people speak faster than they need to, especially under pressure. Slightly slower pace gives your speech production pipeline more time at each stage. Fewer bottlenecks means fewer fillers. You don't need to speak slowly. Just not rushed.

Step 4: Graduate through increasing stakes

Filler reduction is a motor skill. Like any motor skill, it develops best through graduated practice.

Solo narration. Explain things to yourself out loud. Describe your day. Talk through a problem. Without a listener, you can focus purely on the mechanics of pausing instead of filling.

Low-stakes conversation. Practice with friends, family, casual acquaintances. Focus on one thing: replacing "um" with a pause. Don't try to fix "like" and "you know" and "basically" simultaneously. One target at a time.

Structured practice with feedback. This is where tools that track filler words specifically become valuable. Articulated measures filler word usage as one of its seven analysis dimensions -- not just a count, but patterns. When in the conversation did fillers spike? Did they cluster at the start of thoughts or during complex explanations? That granular feedback accelerates the awareness loop because you can see exactly where your brain hits bottlenecks, not just that it hit them.

High-stakes application. Once the replacement habit feels natural in low-pressure settings, bring it into presentations, meetings, interviews. Expect some regression -- stress increases filler rates, and that's normal neurophysiology, not failure. But each time you successfully pause instead of filling under pressure, you strengthen the neural pathway.

Step 5: Accept that this takes time

Filler word patterns are deeply automatic. They've been reinforced through thousands of hours of speaking across your entire life. You're not fixing a setting. You're rewiring a habit.

Most people who practice consistently see noticeable improvement in two to four weeks. Not perfection. Improvement. The filler rate drops, pauses become more natural, self-monitoring becomes less effortful.

Progress isn't linear. You'll have conversations where you sound perfectly clean and others where the fillers come flooding back. Stress, fatigue, unfamiliar topics, and high-stakes situations all temporarily spike your rate. That's expected. It doesn't mean the practice isn't working.

Myths that waste your time

"Intelligent people don't use filler words." False. Research shows zero correlation between intelligence and filler usage. Some studies have actually found that people with larger vocabularies use more fillers -- possibly because more word options create more retrieval competition. Your fillers might be a sign of a bigger vocabulary, not a smaller brain.

"You should never say um." Unrealistic and unnecessary. Zero fillers is not the target. Clear, confident communication where fillers don't distract from your message -- that's the target.

"Fillers mean you don't know what you're talking about." Often the opposite. Fillers frequently spike when speakers are working through complex ideas -- precisely because they know a lot and are trying to express it precisely. An expert navigating nuance will filler more than someone reciting a simple memorized answer.

"Just slow down and the fillers stop." Slowing down helps, but it's one lever among several. Fillers come from multiple mechanisms -- speed is only part of the equation. Awareness, replacement practice, and cognitive load reduction all matter.

What does a realistic goal look like?

If you currently drop a filler every sentence or two, a meaningful target is cutting that rate in half over a month of regular practice. That means:

Recording yourself regularly and studying the patterns
Practicing the pause-instead-of-fill replacement until it's instinctive
Chunking and structuring your thoughts to reduce cognitive load
Being patient when high-stress situations cause temporary spikes

Going from "very noticeable" to "occasionally present" is a massive shift in how people perceive you as a communicator. And it's achievable for virtually anyone willing to practice consistently.

The bottom line

Filler words are not a character defect. They're the natural byproduct of the most complex cognitive process humans perform. Every speaker on Earth uses them.

But in the moments where clear, confident communication matters -- and those moments show up constantly in professional and personal life -- fewer fillers changes how your message lands. Not because the words themselves are harmful. Because their absence creates space. Space for your ideas to breathe. Space for emphasis. Space that sounds like confidence.

Awareness is the starting point. Replacement is the mechanism. Consistent practice is what makes it stick. You don't need to become a different speaker -- just a more articulate one. You just need to give your brain slightly better habits for handling the gaps between thoughts.