Reducing Filler Words Without Going Silent

Everyone who has ever recorded themselves answering an interview question has had the same unpleasant moment: playing it back and hearing um, like, sort of, I guess, basically, you know, scattered through what felt like a fluent answer.

Before you panic about it, some perspective. Fillers are normal features of spoken language. Every assessor uses them. Nobody is counting your ums, and a rubric has no line for them. What they do is act as a symptom, and one particular category of them does real damage. That is what this article is about.

Why you do it

A filler is what comes out when your mouth is running ahead of your planning. You started a sentence before you knew where it ended, and you now need a moment to work it out, so you produce a sound to hold the floor while you catch up.

That tells you what the fix is not. It is not suppressing the sound. Suppress the um without changing the underlying problem and you get a slightly different mess: a stuttered restart, a very long ah, or a silence with a panicked face on top of it. The fix is to stop starting sentences you have not planned.

The fillers that actually cost you marks

Sound fillers, um and ah, are largely cosmetic unless the density gets high enough to be distracting. The ones that genuinely hurt are hedges, because they change the meaning of what you said.

Compare I would speak to my supervisor with I guess I would sort of maybe speak to my supervisor, I think. The action is identical. The second version tells an assessor marking judgement that you do not trust your own reasoning, and that is a domain they are actually recording.

The main offenders to hunt down in your own speech:

• I guess and I think, attached to statements you are actually certain about. Keep them only where you genuinely mean uncertainty.
• Sort of and kind of, which shrink every noun they touch. A kind of difficult conversation is not a difficult conversation.
• Just, as in I would just talk to them. It minimises the exact action you want credit for.
• Obviously and clearly, which are risky in a different way: they imply the assessor should already agree, and they are usually attached to the thing you have not justified.
• To be honest and honestly, which raise a question about the rest of the sentences.
• Trailing tags such as or whatever and things like that, which signal you ran out of thought and hoped nobody noticed.

If you fix nothing else, fix the hedges. They are worth more marks than every um you ever say.

The replacement is a pause, and it has to look like one

The reason candidates fill silence is that from the inside, a pause feels like a failure being broadcast. From the outside, a two second pause is completely unremarkable and often reads as consideration.

The difference between a thoughtful pause and a blank one is almost entirely non verbal. A thoughtful pause has a settled posture, a face doing something deliberate, eye contact that breaks briefly and comes back, and then a clean restart at normal volume. A blank pause has darting eyes, a frozen expression, and a restart that begins with so, yeah.

You can also mark the pause out loud, which removes the ambiguity entirely. Let me think about that for a second. That is a good question, give me a moment. Said calmly, these are signs of composure, not stalling, and they legitimise two or three seconds of quiet.

One practical note on the visual side. Pauses expose your hands. If you are fiddling with a lanyard, a ring, a jacket cuff or a sleeve that will not sit right, the silence fills with visual noise instead. Sorting out what to wear to a medical interview so that nothing needs adjusting is a small thing that makes your pauses look calm rather than anxious.

Drills that actually change the habit

Fillers are a motor habit, so they respond to practice and not to willpower on the day. These are worth doing in the weeks before, ten minutes at a time:

• Plan the sentence, then open your mouth. Deliberately think the first clause through before speaking, rather than starting and steering. This is the root fix and everything else is secondary.
• Shorten your sentences. Long sentences require mid flight planning, which is when fillers appear. Full stops are free.
• Record two minutes daily and count only your hedges, not your ums. Counting the thing that matters trains attention on the right target.
• Practise with a partner who taps the table on every hedge. Immediate feedback rewires this faster than a recording you review an hour later.
• Do a slow motion run. Answer a question at half your normal speed with mandatory pauses at every transition. It sounds absurd and it resets your tolerance for silence.
• Rehearse three transition phrases until they are automatic: the second thing is, what I would actually do is, and the reason that matters is. Prepared transitions occupy the exact moments where fillers breed.

Where fillers are fine, and where they are not

In a roleplay station, perfectly polished speech can actually work against you. Talking to a distressed person with the fluency of a keynote address reads as rehearsed and slightly cold. A natural hesitation while you take in what someone has just told you is human, and humanity is the thing being marked.

The hedges are a different matter even here. If you are meant to deliver a clear message, hedging turns a kind but firm answer into a confusing one, and the actor cannot respond to something you never actually said.

In ethical reasoning and discussion stations, tighten up. Those stations reward a clear position with reasons attached, and hedging language undermines the position while you are still building it. If you want the broader picture of which stations do what, what an MMI actually involves sets out the usual mix.

Do not over correct

There is a failure mode on the other side. Candidates who have been told to eliminate fillers sometimes arrive with speech so controlled it sounds recited, or spend so much attention monitoring their own words that their listening collapses. In a station where you are supposed to be responding to a person, that is a worse trade than the occasional um.

Aim for noticeably fewer, not zero. Nobody in the room is keeping score, and sounding like a person beats sounding like a transcript.

Measure it under pressure

Your filler rate at your desk is not your filler rate in a station, because the whole thing is driven by cognitive load and nerves. You need repetitions with a clock running. MasterMed's live AI interviewer runs timed MMI stations and marks you against a rubric, so you can hear what your speech does under real time pressure and track whether the hedges are dropping across sessions. The first speaking station is free on the trial, no card, and the trial never converts by itself.

One last reframe. Every hedge you delete puts your actual position back in the sentence. That is the real gain here, and it is worth much more than sounding polished: an assessor who can tell exactly what you decided, and why.