"Um" is not a speech defect. It is a placeholder your brain inserts while it searches for the next word, and everyone does it to some degree, including the speakers you admire most. The problem is not that filler words exist. The problem is density: when "um," "like," "you know," and "so" show up every few seconds, listeners stop tracking your content and start tracking your hesitation instead. At a certain frequency, the filler becomes the message.
Most advice on this topic treats fillers as a bad habit to suppress through willpower, which is why most advice does not work. Telling yourself "stop saying um" in the middle of a sentence adds a second cognitive task on top of the one that produced the filler in the first place: word retrieval under time pressure. The fix has to happen upstream, in how you prepare and pace, not in a mid-sentence correction reflex.
Where Fillers Actually Come From
Filler words spike in three specific situations: when you are retrieving a word or fact you have not fully rehearsed, when you are deciding in real time how to phrase something diplomatically, and when you are uncomfortable with silence and rush to fill a pause that would otherwise be productive. Each of these has a different fix, which is why a single generic tip rarely solves the problem for everyone.
If your fillers cluster around specific content, such as technical terms, names, or numbers, the cause is retrieval, and the fix is rehearsal of those specific words until they are automatic. If your fillers cluster around sensitive moments, such as delivering critical feedback or declining a request, the cause is real-time diplomatic editing, and the fix is preparing the phrasing in advance so you are not composing it live. If your fillers are spread evenly throughout your speech regardless of content, the cause is usually discomfort with silence, and the fix is direct: practice letting pauses sit.
The Pause Is the Actual Replacement
The single most effective substitution for a filler word is not another word. It is a silent pause of the same length. A half-second of silence while you retrieve the next phrase reads to an audience as thoughtful and composed. A half-second of "um" reads as uncertain, even though the two moments contain identical information: nothing. The content of the gap has not changed; only your comfort with it has.
This is worth practicing deliberately and separately from your actual content. Record yourself reading a paragraph aloud, and every time you feel the urge to say "um," replace it with a full stop and a breath instead. It will feel unnaturally long to you and sound completely normal to a listener, because your internal sense of pause duration under self-monitoring is reliably longer than reality.
Rehearsal Reduces Fillers More Than Discipline Does
Speakers who rehearse a talk by reading through it silently in their heads consistently produce more fillers when they deliver it than speakers who rehearse out loud, because silent read-throughs do not train the mouth and breath to execute the sentence structures at speaking pace. If a particular transition or explanation reliably triggers a filler for you, the fix is usually to say that specific segment out loud ten or fifteen times until the words come without searching, not to add more general practice time to the whole talk.
This is closely related to building comfort with impromptu speaking: the less time you have had to prepare exact phrasing, the more your brain reaches for filler while it improvises, so building a bank of pre-rehearsed transitions and openers reduces filler load even in unscripted moments.
Slowing Down Buys You Retrieval Time
Nervous speakers tend to speed up, which shrinks the time available for word retrieval and increases filler frequency as a direct consequence. Deliberately slowing your baseline pace by even ten percent gives your brain more processing time per phrase and reduces the pressure that produces filler words in the first place. This is not the same as speaking slowly throughout an entire talk, which can read as low energy. It means building in slightly more space at decision points: right before a name, a number, or a phrase you have not said out loud before.
What Not to Worry About
Occasional fillers in casual conversation are not a communication failure and trying to eliminate them entirely from everyday speech usually produces stiffness that reads worse than the fillers did. The goal for most people is not zero fillers. It is getting the density down in the settings that matter, a presentation, an interview, a pitch, to a level where they stop being noticeable, which for most listeners is somewhere under one filler per thirty to forty seconds of speech. Below that threshold, listeners track your content. Above it, they track your delivery, and your message loses.
Linguists who study spontaneous speech, including work summarized by the Linguistic Society of America, treat filler words as a normal feature of unscripted language production rather than an error, which is a useful reframe: the goal is managing frequency in high-stakes contexts, not eliminating a universal feature of human speech.