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

Abstract

Statistical learning, the process of extracting regularities from the environment, plays an essential role in many aspects of cognition, including speech segmentation and language acquisition. A key component of statistical learning in a linguistic context is the perceptual binding of adjacent individual units (e.g., syllables) into integrated composites (e.g., multisyllabic words). A second, conceptually dissociable component of statistical learning is the memory storage of these integrated representations. Here we examine whether these two dissociable components of statistical learning are differentially impacted by top-down, voluntary attentional resources. Learners' attention was either focused towards or diverted from a speech stream made up of repeating nonsense words. Building on our previous findings, we quantified the online perceptual binding of individual syllables into component words using an EEG-based neural entrainment measure. Following exposure, statistical learning was assessed using offline tests, sensitive to both perceptual binding and memory storage. Neural measures verified that our manipulation of selective attention successfully reduced limited-capacity resources to the speech stream. Diverting attention away from the speech stream did not alter neural entrainment to the component words or post-exposure familiarity ratings, but did impact performance on an indirect reaction-time based memory test. We conclude that theoretically dissociable components of statistically learning are differentially impacted by attention and top-down processing resources. A reduction in attention to the speech stream may impede memory storage of the component words. In contrast, the moment-by-moment perceptual binding of speech regularities can occur even while learners' attention is focused on a demanding concurrent task, and we found no evidence that selective attention modulates this process. These results suggest that learners can acquire basic statistical properties of language without directly focusing on the speech input, potentially opening up previously overlooked opportunities for language learning, particularly in adult learners.

Keywords: Attention; Memory; Neural entrainment; Speech segmentation; Statistical learning.

Figure 1.

Summary of experimental design. The exposure task in the structured condition consisted of 12 min of continuous auditory exposure to four repeating nonsense words. Participants in the Full Attention condition were instructed to attend the speech stream while passively viewing the kaleidoscope images, whereas participants in the Divided Attention condition were instructed to perform a demanding 3-back task on the visual images, while ignoring the speech stream. Our online neural entrainment measure was used to assess statistical learning during the exposure task. After exposure, statistical learning was assessed using the familiarity-rating and target detection tasks.

Figure 2.

Behavioral results reflecting statistical learning, by group. Error bars represent SEM. (A) Performance on the familiarity-rating task. Participants’ ratings did not differ significantly between the two groups. (B) Performance on the target detection task. Participants in the Divided Attention group responded more slowly (top panel) and showed a smaller RT prediction effect (bottom panel), relative to participants in the Full Attention group.

Figure 3.

ITC as a function of frequency and group, computed across the entire exposure period. Shaded regions represent the mean ± SEM. Slightly different stimulus presentation rates were used across participants (see text), and thus the frequencies depicted on the x-axis are expressed relative to the word, second harmonic, and syllabic frequencies used for each individual participant, rather than in numerical terms.

Figure 4.

Modeled progression of ITC at the word and syllable frequency as a function of exposure, based on parameter estimates of fixed effects in the linear mixed-effects model within the first half of exposure. ITC at the word level increased significantly as a function of exposure, while ITC at the syllabic level decreased significantly as a function of exposure, reflecting online learning.

Figure 5.

Partial correlation between the online Word Learning Index (WLI) and the RT prediction effect on the target detection task, controlling for the effect of group (Full versus Divided Attention). The adjusted values shown in this graph represent the residual values for the WLI and RT effect variables after adjusting for the effect of attention group. Greater neural entrainment at the word level relative to the syllable level predicts a larger RT prediction effect.

Figure 6.

ERP results to visual and auditory stimuli, by group. (A) ERP grand averages to images from visual 3-back task. (B) ERP grand averages to individual syllables in the speech stream. (C) ERP grand averages to individual syllables in the speech stream as a function of syllable position, across all participants.