The beat in speech: A window into the attentional mechanisms supporting the detection of non-adjacent dependencies

Converging evidence suggests that musical training can elicit positive transfer effects across multiple domains of language processing, including grammar. In humans, exposure to musical rhythm induces beat and meter perception, which has been shown to enhance attentional allocation and temporal prediction. Theories hypothesize that the predictive gains intrinsic to music rhythmicity may exert cascading effects on syntactic processing by modulating sensitivity to speech prosody. From this perspective, learning should also be boosted insofar as prosody tends to align with grammatical structure. In the present study, we introduce a novel behavioural paradigm to investigate the link between rhythmicity and grammar learning by testing whether the rhythmic beat facilitates the detection of grammar-like structures in artificial languages (ALs), implemented as non-adjacent dependencies (NADs) between variable syllables forming a speech stream (e.g., PU reliably predicts KI in PUlaruKI). A total of 147 participants were exposed to four ALs that varied in rhythmic, grammatical structure, and the alignment between the two: (i) a beat-inducing rhythm with no NADs; (ii) a beat-hindering rhythm with NADs; (iii) a beat-inducing rhythm with embedded NADs temporally misaligned, and (iv) NADs aligned with beat time-points. Results of the implicit and, after exposure, explicit learning measures demonstrate enhanced learning when NADs are embedded within beat-inducing rhythmic structures. Together, these findings suggest that rhythm enhances predictive and attentional mechanisms implicated in grammar learning, underscoring their role in its acquisition.