Startseite Linguistik & Semiotik Learning by predicting: How predictive processing informs language development
Kapitel
Lizenziert
Nicht lizenziert Erfordert eine Authentifizierung

Learning by predicting: How predictive processing informs language development

  • Martin Zettersten
Veröffentlichen auch Sie bei De Gruyter Brill

Abstract

An increasingly influential attempt to provide a unified theory of the mind is grounded in the notion of prediction. On this account, our minds are prediction engines, continuously matching incoming input to top-down expectations. Higher-level predictions or expectations are generated by internal cognitive models at multiple hierarchical levels that jointly serve to minimize prediction error at lower levels in the information processing hierarchy. In language research, prediction has become an increasingly influential approach to understanding how language comprehension unfolds in real time. But how can predictive processing inform our understanding of how we come to learn language in the first place? In this review, I consider how prediction-based theories of the mind can aid in explaining how language development unfolds. First, I review research in perception and language on predictive processes and assess the degree to which they are found in infancy. Next, I consider how predictionbased mechanisms contribute to our understanding of learning, as well as the kinds of patterns that models grounded in prediction can learn. I review research on infants’ prodigious ability to track novel patterns and relate these statistical learning abilities to prediction-based explanations. Finally, I sketch how prediction- based accounts fit within current theoretical positions and debates in the field of language development and suggest directions for future research into how predictive processes support language learning.

Abstract

An increasingly influential attempt to provide a unified theory of the mind is grounded in the notion of prediction. On this account, our minds are prediction engines, continuously matching incoming input to top-down expectations. Higher-level predictions or expectations are generated by internal cognitive models at multiple hierarchical levels that jointly serve to minimize prediction error at lower levels in the information processing hierarchy. In language research, prediction has become an increasingly influential approach to understanding how language comprehension unfolds in real time. But how can predictive processing inform our understanding of how we come to learn language in the first place? In this review, I consider how prediction-based theories of the mind can aid in explaining how language development unfolds. First, I review research in perception and language on predictive processes and assess the degree to which they are found in infancy. Next, I consider how predictionbased mechanisms contribute to our understanding of learning, as well as the kinds of patterns that models grounded in prediction can learn. I review research on infants’ prodigious ability to track novel patterns and relate these statistical learning abilities to prediction-based explanations. Finally, I sketch how prediction- based accounts fit within current theoretical positions and debates in the field of language development and suggest directions for future research into how predictive processes support language learning.

Heruntergeladen am 21.11.2025 von https://www.degruyterbrill.com/document/doi/10.1515/9783110596656-010/html
Button zum nach oben scrollen