Send to

Choose Destination
Front Psychol. 2015 Sep 30;6:1478. doi: 10.3389/fpsyg.2015.01478. eCollection 2015.

Prosodic cues enhance rule learning by changing speech segmentation mechanisms.

Author information

Institució Catalana de Recerca i Estudis Avançats Barcelona, Spain ; Cognition and Brain Plasticity Unit, Hospitalet de Llobregat, Bellvitge Research Biomedical Institute (IDIBELL) Barcelona, Spain ; Department of Basic Psychology, University of Barcelona Barcelona, Spain.
INSERM U955, Equipe 01, Neuropsychologie Interventionnelle, Institut Mondor de Recherche Biomédicale Créteil, France ; Département d'Etudes Cognitives, École Normale Supérieure Paris, France ; Faculté de Médecine, Université Paris-Est Créteil, France ; Assistance Publique-Hôpitaux de Paris, Centre de Référence Maladie de Huntington, Unité de Neurologie Cognitive, Hôpital Henri Mondor-Albert Chenevier Créteil, France.


Prosody has been claimed to have a critical role in the acquisition of grammatical information from speech. The exact mechanisms by which prosodic cues enhance learning are fully unknown. Rules from language often require the extraction of non-adjacent dependencies (e.g., he plays, he sings, he speaks). It has been proposed that pauses enhance learning because they allow computing non-adjacent relations helping word segmentation by removing the need to compute adjacent computations. So far only indirect evidence from behavioral and electrophysiological measures comparing learning effects after exposure to speech with and without pauses support this claim. By recording event-related potentials during the acquisition process of artificial languages with and without pauses between words with embedded non-adjacent rules we provide direct evidence on how the presence of pauses modifies the way speech is processed during learning to enhance segmentation and rule generalization. The electrophysiological results indicate that pauses as short as 25 ms attenuated the N1 component irrespective of whether learning was possible or not. In addition, a P2 enhancement was present only when learning of non-adjacent dependencies was possible. The overall results support the claim that the simple presence of subtle pauses changed the segmentation mechanism used reflected in an exogenously driven N1 component attenuation and improving segmentation at the behavioral level. This effect can be dissociated from the endogenous P2 enhancement that is observed irrespective of the presence of pauses whenever non-adjacent dependencies are learned.


N1; P2; artificial language learning; event-related potentials (ERPs); prosody; rule-learning; segmentation; statistical learning

Supplemental Content

Full text links

Icon for Frontiers Media SA Icon for PubMed Central
Loading ...
Support Center