We often think of fluency in L2 speaking as a matter of speed: fewer pauses, faster delivery, smoother transitions. But what if what you say matters as much as how fast you say it?
That’s what a recent study by Takizawa & Suzuki (2025) suggests. They examined the role of multiword sequences (MWS) – word combinations like agree with, as a result, or in the future – in listener perceptions of fluency. Intriguingly, they found that using more frequent & target-like sequences significantly boosted perceived fluency, even when speakers had identical pause & repair patterns.
The Study Setup
102 Japanese university students gave short argumentative speeches in English about the pros & cons of smartphone use. Their utterances were analysed for:
- Utterance fluency (UF) – speed (e.g., articulation rate), breakdown (e.g., pauses), & repair features.
- Multiword sequences – especially bigrams (e.g., agree with) & trigrams (e.g., as a result of), measured for frequency & sophistication using the COCA corpus [which, with over 1bn words, is probably the most widely-used corpus of English].
- Perceived fluency (PF) – rated by 10 experienced Japanese raters using a 6-point scale after training on specific fluency features.
Importantly, the researchers first looked at how things like pauses & speech speed affected how fluent the speakers sounded — this explained most of the fluency scores (61%). But even after taking those things into account, using more common word combinations (like agree with) still made a small but meaningful difference (0.8%) in how fluent speakers were judged to be.
In short, agree with helped more than agree the – even when the delivery was equally smooth.
Why Does This Matter?
From a psycholinguistic standpoint, MWS reduce cognitive load: they’re stored & retrieved as chunks, not word-by-word. Listeners benefit too – familiar sequences are easier to process (Wray, 2002; Conklin & Schmitt, 2008), leading to more positive fluency judgments.
Ellis (2006) even argues that language users become “probability-tuned” – we expect common phrases & notice when something’s off. This explains why even subtle shifts in phrase use may tip fluency perceptions.
Teacher Takeaways?
- Include MWS in fluency practice: build learners’ repertoire of high-frequency collocations & lexical bundles (as a result, the main reason, on the other hand).
- Teach context-appropriate sequences: Help students notice which chunks commonly appear in argumentative or academic discourse – not just in informal conversation.
- Use corpus-informed materials: Resources based on spoken corpora (e.g., COCA) can highlight what real-world, fluent English sounds like.
Do you ever focus on multiword sequences as part of your fluency work?
Leave a Reply