Do Infants Really Learn Phonetic Categories?

Abstract

Early changes in infants’ ability to perceive native and nonnative speech sound contrasts are typically attributed to their developing knowledge of phonetic categories. We critically examine this hypothesis and argue that there is little direct evidence of category knowledge in infancy. We then propose an alternative account in which infants’ perception changes because they are learning a perceptual space that is appropriate to represent speech, without yet caring up that space into phonetic categories. If correct this new account has substantial implications for understanding early language development.

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

“…This suggests that the critical problem at these ages is learning which dimensions are relevant. This idea is paralleled by several other contemporaneous studies that propose and test similar ideas (Dietrich et al, 2007;Hay et al, 2015) and it is in line with Feldman's (2021) proposal of a reorganizing space. This pushes the developmental window back further -rather than finalizing by 12 months, these studies suggest the developmental window extends at least to 24-month-olds.…”

“…Children acquire speech categories during infancy in a way that can be described as perceptual narrowing (Kuhl et al, 2005;Werker & Curtin, 2005;Werker & Yeung, 2005), thought to be undergirded by a form of unsupervised statistical learning (Maye et al, 2003). However, I argue here that this theory was built on an understanding of the basic nature of speech perception that turned out to be wrong (categorical perception, Liberman et al, 1957), and which enabled an inappropriately rich interpretation of infant methods (Feldman et al, 2021). More importantly, computational modeling suggests a much richer range of potential mechanisms (Nixon & Tomaschek, 2020) and more recent empirical work dramatically alters our understanding of the ecology of the learning environment (Feldman, Myers, et al, 2013;Teinonen et al, 2008) the developmental progression (Hazan & Barrett, 2000;McMurray et al, 2018).…”

“…The apparent limits of distributional learning raise the potential need for supervisory signals that would be much more available in later development. The limits of what can be learned from discrimination methods make it more plausible that infants simply are not acquiring speech categories, but reorganizing the acoustic space (Feldman et al, 2021). And the availability of a wide variety of implicit supervisory signals in the learning ecology (visual speech, articulation, the lexicon) suggests that development could harness more powerful supervisory signals, but these may not be widely available until the second year of life and may be even more powerful later.…”

The Myth of Categorical Perception

“…First, as I described, without the assumptions of CP, it is challenging to conclude that infants truly have acquired speech categories so rapidly. While they are certainly attuning to their language, they may be doing other things (Feldman et al, 2021;McMurray, submitted), and development may be quite a bit slower (Hazan & Barrett, 2000;McMurray et al, 2018) (supporting the Hartshorne et al, 2018, analysis). Second, infant methods are radically different than from those used with adults; a typical infant study may present infants with multiple tokens of the two sounds to be discriminated, and any difference in listening time is used to support discrimination.…”

“…This may require moving beyond categories -which we can't even measure in infancy. For example, Feldman et al (2021) argues that early infancy may be a time in which infants largely organize the perceptual space (e.g., the middle layer of Figure 3), and that true categorization may not come until children are older and have access to more words, articulation, and richer information (see also McMurray, submitted).…”