Abstract
In adults, cognitive fatigue enhances statistical language learning-the ability to detect repeating hidden patterns in continuous speech, and a core component of implicit language acquisition. This supports the cognitive cost hypothesis, which proposes that the adult cognitive architecture, especially executive functions (EFs), constrains effortless language learning. In contrast, articulatory suppression impairs statistical language learning, suggesting the involvement of lower-level auditory-motor mechanisms in acquiring new linguistic knowledge from speech streams. This study examined whether the effects of cognitive fatigue and articulatory suppression on statistical language learning depend on individuals' cognitive resources (CRs). Specifically, we tested whether individual differences in late-developing EFs (1) are associated with statistical language learning ability and (2) modulate the effects of cognitive fatigue and articulatory suppression. Fifty Dutch-speaking first-year university students participated in a multisession statistical-learning experiment. EFs were assessed using three tasks measuring working memory, inhibitory control, and cognitive flexibility. Cognitive fatigue was induced using a time-loaded dual N-back task at participants' maximum speed prior to speech exposure. Articulatory suppression was induced by instructing participants to whisper continuously while listening to the speech stream. Learning was assessed using a post-exposure two-alternative forced-choice recognition task, where participants selected between a pseudoword from the stream and a foil. Based on a factor and cluster analysis of EF scores, participants were grouped into high and low CR groups. In line with the cognitive cost hypothesis, the high CR group performed less well in statistical language learning than the low CR group but benefited from cognitive fatigue. Articulatory suppression impaired statistical learning in the low CR group only, suggesting they rely more on auditory-motor mechanisms. These findings demonstrate that cognitive functioning impacts statistical language learning, suggesting that the ability to acquire new linguistic knowledge from speech streams depends on the interaction between higher-level cognitive and lower-level auditory-motor mechanisms.