You notice it on a recording. Or someone mentions it. Or you just feel it happening in the middle of a sentence: “um,” “uh,” “like,” “you know,” appearing every few words. You decide to stop. You concentrate during the next conversation, and for a while there are fewer of them. Then the topic shifts or the pace speeds up and they are back.
Willpower alone does not fix this. That is because “um” in a second language is not a bad habit. It is a symptom, and the cause is slower word retrieval. That is what needs to change.
Why fillers appear at all
Every speaker, in every language, uses filler words. They serve a real conversational purpose: they signal to the listener that you are still speaking, that you have not finished, that the turn is still yours. Without fillers, a brief pause can be mistaken for a handoff.
Research on the neural activity behind “uh” and “um” confirms this. A 2020 study on the neural dynamics of filler words, published in Scientific Reports, found that these sounds activate speech-related brain areas in ways distinct from meaningless noise. They are part of how spoken conversation stays organized.
The problem is not their existence. It is their frequency. A filler every thirty seconds or so is unremarkable. A filler every few words signals that something in the speech production process is working harder than it should be.
Second language speakers use more fillers
In your first language, most words arrive automatically. You do not consciously reach for “chair” when you want to say “chair.” It simply comes. In a second language, especially at the intermediate level, retrieval is slower. The word is somewhere in memory, but the path to it is less direct.
A 2024 cross-linguistic study in Frontiers in Psychology analyzed filler use across heritage and majority speakers of German, English, and Russian. The study found that bilingual speakers produced significantly more filler particles than monolingual speakers, and that this gap was largest in formal speech: exactly the conditions, job interviews, presentations, group discussions, where you most want to sound fluent.
The extra fillers are not a choice. They are the sound of a brain buying a moment while it completes a retrieval that takes slightly longer than it would in your first language. This is why concentrating harder does not help much: the concentration does not speed up the retrieval. It just makes you more aware of the gap.
What actually shortens retrieval
Retrieval speed improves through repetition. The more times you have retrieved a word in spoken English, the faster the path to it becomes. This is the same process that makes any practiced skill feel automatic: you are not learning something new, you are making something you already know arrive faster.
This means the long-term fix for “um” is the same as the fix for slow speaking generally: more real speaking, with words you actually use, often enough that retrieval becomes less effortful.
One short-term technique helps while retrieval is still developing: replace the filler with a deliberate pause. A silent half-second, a calm breath. This sounds strange when you first do it. But research on the temporal delay hypothesis in speech perception found that brief silences before a word actually help listeners process what is coming next, in much the same way that “um” does, but without drawing attention to itself. A confident pause reads as confidence. An “um” reads as searching.
The limitation of the deliberate pause is that it is a cover. You are pausing instead of filling, but the retrieval speed itself has not changed. Over months of speaking practice it will. But that requires speaking often enough that the words you most often reach for start showing up without a delay.
Where to get the repetitions
The words that cause “um” are usually not the common ones. They are the specific term, the right connector between two ideas, the precise phrase for a feeling. You already retrieve everyday words fast. The slower ones are the ones you do not reach for as often.
These get faster through conversations where you are pushed to find them. Solo practice has a real limit here: when you choose your own topic and your own direction, you tend to use the words you already retrieve quickly. The uncomfortable stretch, finding the right phrase for an idea you did not plan to express, happens more in conversations you did not control.
When a discussion partner asks you something you did not anticipate, or takes a topic in a direction you were not expecting, you have to retrieve whatever fits right now. Successful or not, those retrieval attempts are what build the speed.
PractE’s live group discussions create exactly that pressure: structured turns, unscripted responses from other speakers, and topics you did not choose. Each turn pushes retrieval in real conditions. The gap between “um” and the word arriving closes through exactly those moments, and through building the habit of speaking more often until the most common paths become reflex.
