CoChat

The DOI matches its Crossref record; 46 of the 46 references it cites were found in a registry or an index; none of the 39 sources its claims cite is checked for support yet.

Statistical learning of speech regularities can occur outside the focus of attention

Open access

Overview

CoChat Summary

TL;DR

Statistical learning of speech regularities was examined under full versus divided attention using EEG neural entrainment and offline tests. Diverting attention did not alter online perceptual binding of syllables into words or familiarity ratings, but reduced an indirect reaction-time-based memory measure, suggesting attention differentially affects memory storage rather than perceptual binding.

Key findings

  • Neural entrainment to component words did not differ between Full and Divided Attention groups (ITCword: F(1,72) = 0.067, p = 0.80), whereas entrainment to syllables was larger in Full Attention (F(1,54) = 5.96, p = 0.018).
  • Across participants, ITCword increased and ITCsyllable decreased with exposure (F(1,2481) = 9.89, p = 0.002; F(1,2478) = 6.74, p = 0.009), mainly in the first half of exposure (~6 min), with no group difference in the ITCword time course (interaction F(1,1282) = 0.30, p = 0.59).
  • On target detection, Full Attention showed a larger RT prediction effect than Divided Attention (18.3% vs. 10.9% RT decrease from Position 1 to 3; t(44) = −2.16, p = 0.038); WLI and Group independently predicted the RT prediction effect (F(2,43) = 5.40, p = 0.008; Group t(43) = 2.84, p = 0.007; WLI t(43) = 2.31, p = 0.026).
  • Familiarity ratings showed learning across groups (Word Category F(2,88) = 31.4, p < 0.001; linear effect F(1,44) = 66.3, p < 0.001) but no significant group difference (Group × Word Category F(2,88) = 0.90, p = 0.41; familiarity score Full mean = 0.79, SEM = 0.56 vs. Divided mean = 0.55, SEM = 0.60; t(44) = 1.38, p = 0.18). Manipulation checks: P300 to visual images was larger in Divided Attention (F(1,44) = 61.74, p < 0.001); N100 to syllables was larger in Full Attention (F(1,44) = 4.27, p = 0.045).

Methods

  • 49 participants (mean age = 21.8 years, SD = 2.8; 37 female/12 male) were randomly assigned to Divided Attention (n = 25) or Full Attention (n = 21); 3 additional Full Attention participants were excluded for EEG technical failure (n = 2) or <10% target detection (n = 1).
  • Exposure used a continuous speech stream of 4 repeating trisyllabic nonsense words, 2400 syllables/800 words over 3 blocks (~4 min each), with within-word transitional probability 1.0 vs. between-word 0.33; the Divided group performed a concurrent visual 3-back task on kaleidoscope images, while the Full group viewed images passively. This was followed by a familiarity-rating task (12 trials: 4 words, 4 part-words, 4 non-words; 1–4 scale) and a target detection task (36 streams; 48 trials per syllable position).
  • EEG was recorded at 512 Hz from 64 Ag/AgCl electrodes; neural entrainment was quantified as inter-trial coherence (ITC) at word and syllable frequencies, with Word Learning Index = ITCword/ITCsyllable; ERPs included P300 to images and N100 to syllables. Analyses used repeated-measures ANOVAs, robust linear mixed-effects models, and regressions.

Limitations

  • Stated: attention cannot be completely abolished; Divided Attention participants may have alternated attention to the speech stream, and residual attention/working memory may have supported learning. The familiarity-rating task may have been insufficiently sensitive to reveal group differences, and the sample size may have limited detection of explicit memory differences.
  • Stated: the artificial speech segmentation paradigm may not scale to real-world language learning; future work should test within-modality auditory or linguistic competing tasks.
  • Evident from the text: stimulus presentation rate changed partway through data collection due to a hardware upgrade (260–300 ms/syllable), though balanced across groups and not significantly different (t(44) = 0.033, p = 0.97); the study used a cross-modality visual distractor, so conclusions about attention may not generalize to auditory/linguistic competition.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.cortex.2019.01.013

Cites

46 of 46 references verified0 citation errors · 0 not checkable

Statistics

36 tests recomputed · 0 with the reported and recomputed p on opposite sides of .05

Verification

Claims the paper makes

0 Valid · 1 Not valid · 34 Not checked · 106 Own results and interpretations

72 more claims

Create a free account to see all 141 claims and to verify the 29 references still open.

See all claims — free account
  • Probably not a substantial claim. 31
  • It may state two separate claims as one. 41
  • The quote only partly states this claim. 3

Create a free account to see these claims.

Statistics that don’t add up

Each reported p-value recomputed from its test statistic and degrees of freedom.

1 of 1

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Doesn’t add upt(44) = -2.16reported p = .038 · recomputed p = .036

    “t(44) = −2.16, p = .038”

  • Consistentt(44) = 0.033reported p = .97 · recomputed p = .974

    “t(44) = 0.033, p = .97”

  • ConsistentF(2, 88) = 31.4reported p < .001 · recomputed p < .001

    “F(2,88) = 31.4, p < .001”

  • ConsistentF(1, 44) = 66.3reported p < .001 · recomputed p < .001

    “F(1,44) = 66.3, p < .001”

  • ConsistentF(2, 88) = 0.9reported p = .41 · recomputed p = .410

    “F(2,88) = 0.90, p = .41”

5 of 36 · the rest aren’t loaded

31 more recomputed tests

Create a free account to see every test the Stats check recomputed.

See all tests — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 46 references

Research works printed but not linked: 2

  1. Lehiste, 1960, An acoustic phonetic study of open juncture, Phonetica Supplementum, 5, 1
  2. Treisman, 1973, Divided attention between ear and eye, Vol. 4, 101

Cited by

See all 111

Metadata

Authors
Batterink, L. J., & Paller, K. A.
Affiliations
Northwestern University (US)
Corresponding author
Batterink, Laura J.
Venue
Cortex
Publisher
Elsevier BV
Published
2019-06
Volume
115
Pages
56-71
ISSN
0010-9452 (ISSN-L)
OpenAlex
W2911556349
Cited by
111
References
46
Citations by year
2026: 25 · 2025: 19 · 2024: 12 · 2023: 19 · 2022: 9 · 2021: 16 · 2020: 10 · 2019: 1
FWCI
4.16
Type
journal article
Language
en
Open access
Locations
  • Cortex (acceptedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
  • Scholarship@Western (Western University) (submittedVersion)
Funders
  • National Institute of Neurological Disorders and Stroke (T32 NS047987)
  • National Science Foundation (BCS-1461088)
  • National Institute of Child Health and Human Development (F32 HD078223)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neurobiology of Language and Bilingualism
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Phonetics and Phonology Research
  • Social Sciences › Psychology › Developmental and Educational Psychology › Language Development and Disorders
Keywords
statistical learning, speech segmentation, selective attention, top-down processing, perceptual binding, neural entrainment, EEG, language acquisition, adult language learning, polysyllabic words
MeSH terms
Acoustic Stimulation, Attention, Electroencephalography, Evoked Potentials, Female, Humans, Learning, Male, Memory, Reaction Time, Speech, Young Adult

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-19, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-19 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.