The acquisition of speech categories: Beyond perceptual narrowing, beyond unsupervised learning and beyond infancy.

Abstract

An early achievement in language acquisition is carving up a variable acoustic space into linguistic categories. The canonical story is that infants accomplish this in the first years, when there are few avenues for learning other than unsupervised statistical learning. This review synthesizes psychological, phonetic and computational results that challenge this view. First, unsupervised learning may be insufficient given the statistics of speech (including infant directed speech). Second, the evidence that infants “have” speech categories is not unambiguous. Third, error-driven/discriminative learning can help even in unsupervised learning environments. Fourth, implicit supervisory signals are available to older infants. Finally, development is protracted through adolescence, enabling richer avenues for development. Thus, infancy may be a time of organizing the auditory space with true categorization arising in a complex developmental cascade later in life. This has implications for critical periods, second language acquisition, and our basic framing of the problem of speech perception.

Learning to unlearn: The role of negative evidence in morphophonological learning

“…Although overshadowing occurs across the whole learning process, children eventually overcome these differences and learn the appropriate form-meaning mappings. How the language input within the environment -the learning ecology -interacts with the learning mechanisms themselves can shed lights on how the learners achieve this feat (see McMurray (2022) for an overview on the difference between ecology and mechanism).…”