Explicit Instructions Do Not Enhance Auditory Statistical Learning in Children With Developmental Language Disorder: Evidence From Event-Related Potentials

Abstract

A current issue in psycholinguistic research is whether the language difficulties exhibited by children with developmental language disorder [DLD, previously labeled specific language impairment (SLI)] are due to deficits in their abilities to pick up patterns in the sensory environment, an ability known as statistical learning (SL), and the extent to which explicit learning mechanisms can be used to compensate for those deficits. Studies designed to test the compensatory role of explicit learning mechanisms in children with DLD are, however, scarce, and the few conducted so far have led to inconsistent results. This work aimed to provide new insights into the role that explicit learning mechanisms might play on implicit learning deficits in children with DLD by resorting to a new approach. This approach involved not only the collection of event-related potentials (ERPs), while preschool children with DLD [relative to typical language developmental (TLD) controls] were exposed to a continuous auditory stream made of the repetition of three-syllable nonsense words but, importantly, the collection of ERPs when the same children performed analogous versions of the same auditory SL task first under incidental (implicit) and afterward under intentional (explicit) conditions. In each of these tasks, the level of predictability of the three-syllable nonsense words embedded in the speech streams was also manipulated (high vs. low) to mimic natural languages closely. At the end of both tasks’ exposure phase, children performed a two-alternative forced-choice (2-AFC) task from which behavioral evidence of SL was obtained. Results from the 2-AFC tasks failed to show reliable signs of SL in both groups of children. The ERPs data showed, however, significant modulations in the N100 and N400 components, taken as neural signatures of word segmentation in the brain, even though a detailed analysis of the neural responses revealed that only children from the TLD group seem to have taken advantage of the previous knowledge to enhance SL functioning. These results suggest that children with DLD showed deficits both in implicit and explicit learning mechanisms, casting doubts on the efficiency of the interventions relying on explicit instructions to help children with DLD to overcome their language difficulties.

As trials go by: Effects of 2-AFC item repetition on statistical learning performance.

“…We acknowledge that this problem is not easy to solve, as the use of a low number of observations also raises important reliability problems, and the use of other postlearning tasks such as 3- or 4-AFC tasks, pattern completion tasks, target detection tasks, or even the induced chunking recall task, necessarily involves exposing participants to further stimulation, thereby allowing for more room for interference to occur. Researchers should, therefore, look for other strategies to provide a better proxy for SL, such as collecting electrophysiological data as the exposure to the auditory stream unfolds, a method being increasingly used in SL research (e.g., Abla et al, 2008; Cunillera et al, 2006; De Diego Balaguer et al, 2007; Gutiérrez-Domínguez et al, 2022; Soares, Gutiérrez-Domínguez, et al, 2020; Soares, Gutiérrez-Domínguez, Lages, et al, 2022; Soares, Gutiérrez-Domínguez, Oliveira, et al, 2022; Soares, Lages, et al, 2022; Soares, Lages, Oliveira, & Gutiérrez-Domínguez, 2021), or using behavioural tasks such as a self-paced statistical learning task (e.g., Siegelman et al, 2018; Siegelman, Bogaerts, Armstrong, & Frost, 2019), asking participants to press a key to proceed to the following stimulus during exposure, while using differences between responses times to predictable (second and third elements) versus unpredictable stimuli (first element) within a triplet as the exposure unfolds as a better proxy of SL.…”

“…Specifically, participants were told they would listen to “new words” from another foreign language. Each of the eight new “words” was then auditorily presented, and participants were asked to repeat each of them correctly before the familiarization phase began (see Batterink, Reber, & Paller, 2015; Gutiérrez-Domínguez et al, 2022; Lages et al, 2022; Soares, Gutiérrez-Domínguez, Lages, et al, 2022; Soares, Gutiérrez-Domínguez, Oliveira, et al, 2022; Soares, Lages, Oliveira, & Gutiérrez-Domínguez, 2021, for a similar procedure). Like in the implicit task, during the familiarization phase participants were asked to press a button whenever they heard the click sound, to ensure they paid attention to the auditory stimuli.…”

“…Even though no previous study has attempted to directly address these issues with the collection of “real” data from “real” participants, we anticipate that increasing the number of 2-AFC repetitions will contribute to hamper “word” discrimination, as reflected by a significant decrease in the proportion of correct responses as the 2-AFC task unfolds or, in other words, as 2-AFC item repetition increases. Stronger item repetition effects are also expected for “words” with low- rather than high-TPs since “words” with higher levels of predictability are expected to produce stronger representations (e.g., Siegelman et al, 2018; Soares, Gutiérrez-Domínguez, et al, 2020; Soares, Gutiérrez-Domínguez, Lages, et al, 2022; Soares, Gutiérrez-Domínguez, Oliveira, et al, 2022). This might make these words less prone to interference, particularly when the 2-AFC task is performed under implicit conditions.…”

“…The use of a small number of 2-AFC trials (e.g., four in Spencer et al’s, 2014 work) and of triplets with the same level of predictability (i.e., TPs of 1.0) certainly contribute to this state of affairs, as they do not allow for enough variance either within the sample or along with the ability to assess SL properly (see Arnon, 2020; Erickson et al, 2016; Frost et al, 2019; Siegelman, Bogaerts, Armstrong, & Frost, 2019; Siegelman, Bogaerts, Christiansen, & Frost, 2017; Siegelman, Bogaerts, & Frost, 2019; van Witteloostuijn et al, 2019, for similar arguments). To overcome such flaws, researchers have adopted different solutions, such as (a) increasing the number of 2-AFC trials by contrasting the triplets presented during exposure against different foils (e.g., 16 times in Arciuli & Simpson’s, 2012 study); (b) testing not only triplets but also pairs (bigrams) of stimuli (e.g., Siegelman et al, 2018; Siegelman, Bogaerts, & Frost, 2017); (c) using other variants of the 2-AFC task, such as a four-alternative forced-choice (4-AFC) pattern recognition task (e.g., Bertels et al, 2015; Siegelman et al, 2018; Siegelman, Bogaerts, & Frost, 2017) or a three-alternative forced-choice (3-AFC) completion task (e.g., Bertels et al, 2012, 2015; Siegelman, Bogaerts, & Frost, 2017; van Witteloostuijn et al, 2019); (d) using other experimental tasks, such as a target detection postlearning task (e.g., Batterink, Reber, & Paller, 2015; Batterink, Reber, Neville, & Paller, 2015; Siegelman et al, 2018; van Witteloostuijn et al, 2019), a statistically induced chunking recall task (e.g., Isbilen et al, 2020; Kidd et al, 2020), or a self-paced statistical learning task (e.g., Siegelman et al, 2018; Siegelman, Bogaerts, Armstrong, & Frost, 2019; van Witteloostuijn et al, 2019); or even (e) collecting event-related potentials (ERPs) data during the test and/or familiarization phases (e.g., Abla et al, 2008; Batterink, Reber, & Paller, 2015; Batterink, Reber, Neville, & Paller, 2015; Cunillera et al, 2006; De Diego Balaguer et al, 2007; Gutiérrez-Domínguez et al, 2022; Sanders et al, 2002; Soares, Gutiérrez-Domínguez, et al, 2020; Soares, Gutiérrez-Domínguez, Lages, et al, 2022; Soares, Gutiérrez-Domínguez, Oliveira, et al, 2022; Soares, Lages, et al, 2022; Soares, Lages, Oliveira, & Gutiérrez-Domínguez, 2021).…”

“…These results led the authors to conclude that neither explicit nor implicit representations are impervious to interference, even though the between-subjects design adopted in that study, as well as in many other studies exploring the type of representations emerging from SL tasks (e.g., Arciuli et al, 2014; Batterink, Reber, Neville, & Paller, 2015; Batterink, Reber, & Paller, 2015), prevents us from drawing strong conclusions, since previous studies found substantial variability in the way individuals respond to SL tasks (see Siegelman, Bogaerts, Christiansen, & Frost, 2017; and also Siegelman, Bogaerts, & Frost, 2017), which might have confounded previous results. The use of within-subjects designs is thus strongly advisable (see Soares, Gutiérrez-Domínguez, et al, 2020; Soares, Gutiérrez-Domínguez, Lages, et al, 2022; Soares, Gutiérrez-Domínguez, Oliveira, et al, 2022, for recent examples), even if it necessarily implies not counterbalancing the order of the SL tasks across participants. Note that once an SL task is performed explicitly by participants, it can no longer be performed implicitly by the same participants, hence making the implicit followed by the explicit presentation of the SL task the only viable solution in that type of design.…”