If you’ve ever tried to sound fluent in another language, you’ve probably focused on the obvious things: fewer pauses, smoother sentences, and a decent accent. But what actually makes a listener think you’re fluent?

According to a new study by Takizawa and Suzuki (2025), it’s not just about speed or avoiding “uhm” and “ehh.” It turns out that using the right word combinations—multiword sequences like “on the other hand” or “as a matter of fact”—has a small but consistent effect on how fluent you sound.

The researchers worked with 102 Japanese university students who gave argumentative speeches in English. A group of trained raters judged their fluency, while the researchers measured their actual speech characteristics (pause length and speech rate) and how often they used common word combinations found in English corpora.

Here’s the interesting part: even after controlling for all those typical fluency markers—how fast someone spoke, how many times they paused, whether they corrected themselves mid-sentence—just using more frequent multiword sequences still had a statistically significant effect on how fluent the speaker was perceived. It’s not a huge effect, but it’s measurable.

So, what does that mean for learners? It implies that sounding fluent isn’t only about speaking quickly or smoothly. It’s also about sounding natural. That includes picking up on and using the chunks of language that native speakers always rely on. Think “I think it’s important to…” or “the main reason is…” These aren’t flashy phrases, but they help your speech flow in a way that listeners intuitively recognize as fluent.

In other words, don’t just speak well if you want to look fluent. Speak predictably. Use the patterns your listener expects. That’s not boring—it’s smart. It makes your speech easier to process and easier to follow. As this study shows, this makes you sound more fluent—even if you still say “uhm” occasionally.

Abstract of the study:

This study explored how second language (L2) speakers’ use of multiword sequences in speech predicted perceived fluency ratings while controlling for their utterance fluency. A total of 102 Japanese speakers of English delivered an argumentative speech, which was analyzed for bigram and trigram measures (frequency, proportion, and mutual information) and utterance fluency measures capturing three subdimensions: speed, breakdown, and repair fluency (Tavakoli & Skehan, 2005). Perceived fluency was assessed by 10 experienced L2 raters. Mixed-effects regression analyses revealed that after establishing the parsimonious model solely by UF predictors (marginal R2 = .61), a frequency-based n-gram predictor––bigram proportion––slightly but significantly explained the remaining variance of fluency rating scores (0.8%). The results indicated that multiword sequences in speech had a small but systematic impact on perceived fluency, even controlling for the effects of utterance fluency. This finding contributes to the discussion concerning the role of multiword sequences in fluent speech production.