Filler Words Blog Header copy

Have you been told to use fewer filler words? Do you feel self-conscious about your filler words when you read transcripts or listen to recordings of yourself speaking? You may be worrying needlessly: filler words can sometimes be friends.

What are filler words?

Every language has filler words; we use them every day in conversation. Some English examples are:

I’d love to go to the party tomorrow, but I, um, already have plans.1

I want to make sure we’re, you know, on the same page and everything.

When you have work to do, it’s not always easy to, like, find the motivation to sit down and do it.

Google “filler words,” and you’ll be bombarded with blog posts, listicles, and editorials; none of them positive. We love to hate filler words.

And it’s true: we don’t want to hear them in speeches, scripts, or research papers. But, as a linguist, I hope you do use them in conversations. To understand why, let’s take a brief dive into some basic linguistics.

Form and function

You may have noticed that some of the examples above are words that also do other things in English—they only moonlight as filler words. For example, the word like does a lot of overtime:

I like cats. (verb)

That dog looks like a cloud. (preposition)

Like, the building is really old, is what I’m getting at. (particle)

My dog is, like, super needy. (discourse-marker)

And I was like, “Why would you do that?” (quotative)

This word has one form (like), but many functions. Linguists call this form-function asymmetry, and it happens in every language. All it means is that a given word or phrase can serve different purposes in different contexts.

So, obviously we can’t get rid of like entirely, because it serves many functions.

Filled pauses

The function that like and its ilk serve even in their capacity as filler words is also important.

Let’s return to this sentence:

My dog is, like, super needy.

Here, like’s function is to be a filled pause. Filled pauses are integral to spoken conversation: using one holds the floor for you to express yourself clearly and correctly. It usually tells the listener: I’m not finished my thought, hold on just a minute while I express it in the clearest way I can think of.

A filled pause can also emphasize the emotion or severity of what we’re about to say. For example:

Um, did you really just say that to me?

Here, the speaker is shocked and/or angry about what someone has just said. The um indicates that what was said was so inappropriate that the speaker needs a beat to process it.

Backchannels and affirmatives

Another type of filler word, the backchannel, is used mainly in interactional contexts. A backchannel’s function is to be used by the listener rather than the speaker, and it usually lets the speaker know: I’m paying attention to what you’ve said so far; please continue. For example:

Speaker 1: So what he said to me was that they were trying to find a distributor for the materials, someone who could get those out to us quickly, and then we would be able to create and ship the order.

Speaker 2: Uh-huh.

Speaker 1: So once we can find a distributor, we'll know how long it will take to deliver the materials, and we’ll let you know when the order would ship based on that info. Does that work for you?

Speaker 2: Yes. Thank you, please do keep me in the loop.

The backchannel uh-huh conveys important social and linguistic information to Speaker 1: it says that Speaker 2 understood them, and doesn’t want a turn speaking yet.

As if that weren’t enough, form-function asymmetry strikes again: Speaker 2 could have used uh-huh in their final utterance too:

Speaker 2: Uh-huh. Thank you, please do keep me in the loop.

This function of uh-huh isn’t exactly rare, either: in our own transcript data, at least 30% of the time a backchannel like uh-huh appeared in a transcript, it was actually being used to reply to a question in the affirmative. Of those that are used as affirmative responses, up to 65% of these occur with no other affirmative word in the preceding or following sentence—meaning that these words are the only indication that the speaker has responded “yes.”

People sometimes expect filler words

If you’re still on the fence about fillers, here’s another surprising fact: in some contexts, not using filler words actually makes people uncomfortable.

A major struggle faced by individuals on the Autism Spectrum is having verbal interactions with others—but it can be difficult to pinpoint what exactly is “awkward” or “different” about their speech. Recent findings provide a hint about what’s going on: people with autism spectrum disorder (ASD) tend to use fewer filler words in conversations than neurotypicals do. (This difference disappears when those with ASD are talking to someone they know is actually interested in having the conversation.) That is, not using enough filler words in a conversation makes the interaction sound strange to neurotypicals—even though we don’t consciously realize that’s what’s going on.

Why? The answer lies in the nature of a conversation: it’s interactional. That means we (usually) need to take conversational turns, we (usually) want people to understand what we’re saying, and we (usually) want them to feel good about their interactions with us. Filler words allow us to do all those things, and to do them without a lot of conscious thought.

That means that there can be consequences to not using filler words. For example, potential clients tend to think of a sales call (learn more about sales call reporting) as a conversation about your product meeting their needs. That means they’re expecting filler words; without them, the interaction might be interpreted as scripted or overly rehearsed. These words—be they filled pauses, backchannels, or other discourse markers—are part of what we expect in a conversational context. Remove them, and you’ve changed the context to a more formal, less interactional one.

So if you’re a conversational filler word fanatic, then like, just keep talking—or filling those pauses—the way you want.

Bio: Shayna is an NLP Engineer at Dialpad, where she analyzes everyday language as it relates to Dialpad Ai. She received her PhD in Linguistics from the University of Toronto in 2017, with a focus on sociolinguistics and morphosyntax.


1 All of the examples present in this post are fabricated, but are intended to represent natural language. We respect our clients’ privacy.

References & Further Reading