We’re all familiar with the scenario: you’re in a café or on public transport, half-listening to the buzz of voices around you. You don’t catch any actual words, yet you still know when someone is speaking Spanish, Arabic or German. Something in the rhythm & sound patterns gives it away almost instantly. That ability to recognise a language without understanding it reflects a kind of knowledge we rarely stop to think about. A recent open-access study (2025) in Frontiers in Language Sciences zooms in on this hidden layer, asking whether English learners have an intuitive sense of what “sounds like English”, even when the word itself does not exist.
The study
The researchers investigated whether adult learners of English show sensitivity to English phonotactics, the patterns that determine which sound sequences are more or less word-like.
Participants were native speakers of Arabic (Saudi Arabia) & Mandarin (China), divided into Beginner & Advanced groups using two different proficiency metrics: – number of university English semesters (the Arabic speakers), & scores on the CET-4 standardised test (the Mandarin speakers).
Learners completed a word-likeness rating task. They listened to 60 English-like non-words (e.g. items that followed English sound patterns but had no meaning) & rated each on a 1–7 scale from “good English word” to “bad English word”.
Crucially, the non-words varied in phonotactic probability, calculated using established computational tools that measure how frequent particular sounds & sound sequences are in real English.
The researchers then compared learners’ subjective ratings with these objective phonotactic measures, as well as with ratings from native English speakers reported in earlier work (Vitevitch & Donoso, 2012).
The findings
Across both language groups & proficiency levels, learners’ intuitions aligned strongly with English phonotactic patterns.
Correlations between learners’ ratings & phonotactic probability were statistically significant for all groups, with medium to large effect sizes. In other words, learners reliably judged which fake words sounded “more English”.
Surprisingly, Beginners & Advanced learners did not differ significantly from each other in how strongly their ratings tracked phonotactic probability.
However, both groups differed from native speakers. Learners tended to rate non-words as more word-like overall, especially Mandarin speakers, whose mean ratings were significantly higher than those of native speakers.
Learners’ ratings correlated more strongly with each other than with native speakers’ ratings, suggesting that classroom experience quickly reshapes phonotactic intuition, but reaching fully native-like sensitivity may require much broader lexical exposure.
This fits neatly with earlier work showing that phonotactic sensitivity emerges early in infancy (Jusczyk et al., 1993; Sundara et al., 2022), influences word learning & segmentation across the lifespan, & relates to pronunciation quality in L2 learners (Kivistö-de Souza, 2017).
In other words, a learner might instantly feel that splent could be an English word, while bnika feels wrong, even if neither has ever appeared in a textbook. That intuition reflects real statistical knowledge of English, acquired implicitly.
Why it matters
The study reinforces a growing idea in SLA: learners acquire rich phonological knowledge implicitly, often ahead of what traditional tests capture.
It also highlights the potential of non-word tasks as low-cost, low-burden ways of probing phonological proficiency, without relying on reaction times, trained raters or production accuracy.
Teacher Takeaways?
There’s no direct classroom application for all this, but a few thoughts come to mind:
- Learners’ “gut feelings” about sound patterns are meaningful data, not noise.
- Phonotactic awareness develops early & may plateau before other skills visibly improve.
- Exposure to a wide lexical range may be key to moving from learner-like to native-like intuitions.
For language educators, this research is fascinating because it reminds us that progress is not always linear or visible. Learners may already “sound English” internally long before their output reflects it.
What kinds of learner knowledge do you think our usual assessments fail to capture?
Leave a Reply