Does second-language skill help you hear speech in noise?
Swedish adults with better English understood English sentences in noisy, talker-filled backgrounds much better, while their English level made no difference for Swedish sentences.
Source
The influence of non-native language proficiency on speech perception performance
Study at a glance
- Design
- Human experiment — Within-subject speech-in-noise test: target language (Swedish, English) x four maskers (stationary noise, fluctuating noise, Swedish babble, English babble), plus proficiency, reading span and Raven measures.
- N
- N=23 · 23 native Swedish adults; for a follow-up analysis they were split at the median English score into low and high proficiency subgroups, excluding those exactly at the median.
- Population
- Normal-hearing native Swedish adults aged 28-64 recruited from workplaces, with English as a second language.
- Outcome
- Speech reception threshold (the signal-to-noise ratio needed to repeat half of sentences correctly).
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Twenty-three native Swedish adults with normal hearing repeated everyday sentences in Swedish and in English while hearing one of four background sounds: steady noise, fluctuating noise, or two people talking in Swedish or in English. The noise level was adjusted adaptively to find the level at which each person got half the sentences right. Participants also took a standardised English comprehension test, a reading span test of working memory in both languages, and Raven's matrices.
What they found
Everyone needed a quieter background to understand English than Swedish, and competing talkers interfered more than noise. People with higher English proficiency had better thresholds for English sentences but not for Swedish ones, a strong language-by-proficiency interaction. Raven's non-verbal reasoning was unrelated to any threshold, and working memory measured in English related to English thresholds, which the authors attribute mainly to English skill rather than memory capacity.
The limits
What it doesn't show
The sample was small and many correlations were computed, so the authors themselves urge caution; one unexpected correlation was driven by a single listener. Proficiency was measured, not manipulated, so it may stand in for related factors such as how often people use English, and it was linked with age. The results come from one group of Swedish speakers with English as their second language, and they do not settle whether working memory matters for non-native listening, since the study may have been too small to detect it.
Key terms
- Speech reception threshold (SRT)
- The signal-to-noise ratio at which a listener correctly repeats half of the sentences; lower is better.
- Energetic masking
- Interference caused by background sound physically overlapping the target speech in frequency and time, like steady noise.
- Informational masking
- Extra interference from background sound that carries meaning, such as understandable talkers, which competes for attention and processing.
- Reading span test
- A working memory task where people judge sentences while remembering words from them, measuring storage plus processing.
- Ease of Language Understanding (ELU) model
- A model proposing that working memory and knowledge are recruited more when speech is hard to match to stored representations.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What does a lower speech reception threshold indicate?
Common questions
Why did babble interfere more than noise?
Two people talking carry words and meaning that compete with the target sentence for attention, adding informational masking on top of simple acoustic overlap.
Is working memory unimportant for non-native listening?
The study found no strong link, but it had few participants; the English reading span test seemed to reflect English skill as much as memory, so the question stays open.
Why test adults of many ages instead of students?
A wider age and education range gave more variation in proficiency and performance, making it easier to detect individual differences and more representative of the general population.
More on Language