Distributional learning has immediate and long-lasting effects Paola Escudero a,⇑ , Daniel Williams b a MARCS Institute, University of Western Sydney, Bankstown Campus, Locked Bag 1797, Penrith, NSW 2751, Australia b Institute of Humanities & Creative Arts, University of Worcester, Henwick Grove, Worcester WR2 6AJ, United Kingdom article info Article history: Received 25 July 2013 Revised 3 July 2014 Accepted 8 July 2014 Keywords: Statistical learning Distributional learning Longitudinal development Non-native sound discrimination abstract Evidence of distributional learning, a statistical learning mechanism centered on relative frequency of exposure to different tokens, has mainly come from short-term learning and therefore does not ostensibly address the development of important learning pro- cesses. The present longitudinal study examines both short- and long-term effects of dis- tributional learning of phonetic categories on non-native sound discrimination over a 12-month period. Two groups of listeners were exposed to a two-minute distribution of auditory stimuli in which the most frequently presented tokens either approximated or exaggerated the natural production of the speech sounds, whereas a control group listened to a piece of classical music for the same length of time. Discrimination by listeners in the two distribution groups improved immediately after the short exposure, replicating previ- ous results. Crucially, this improvement was maintained after six and 12 months, demon- strating that distributional learning has long-lasting effects. Ó 2014 Elsevier B.V. All rights reserved. 1. Introduction It is well known that young infants are able to discrim- inate virtually all speech sounds and that this ability declines as they become more attuned to their native lan- guage. By the end of their first year, infants’ discrimination of speech sounds resembles more closely that of adults of the same native language. While the change in infants’ speech perception may be related to or at least coincide with the onset of word learning, word learning alone can- not account for the change because speech perception becomes specific to the infants’ native language before they can reliably distinguish words. Infants’ learning of speech sound contrasts is likely to be underpinned by their sensitivity to the statistical regularity of the auditory input in their ambient spoken language (for a review, see Krogh, Vlach, and Johnson, 2013). In that respect, infants have been shown to harness relative frequency distributions in the continuous auditory signal to learn to discriminate speech sounds, a domain-general mechanism known as distributional learning (Maye, Weiss, & Aslin, 2008; Maye, Werker, & Gerken, 2002; Werker, Yeung, & Yoshida, 2012; Yoshida, Pons, Maye, & Werker, 2010). In distributional learning experiments, listeners are presented with auditory stimuli that form a continuum and vary in equal steps along a particular acoustic-pho- netic dimension. The stimuli are either presented with fre- quencies that constitute a bimodal distribution, in which tokens near the endpoints of the continuum are most fre- quent, or a unimodal distribution, in which tokens around the middle are most frequent. A bimodal distribution mim- ics how the speech sounds appear in a binary sound con- trast in a language because they tend to be produced with properties that place them near the edges of an acous- tic-phonetic continuum. For instance, Maye et al. (2002) used the contrast ‘‘da’’ and ‘‘ta’’, where ‘‘t’’ did not have the aspiration of English ‘‘t’’ but mimicked how ‘‘t’’ as pro- nounced in languages such as Spanish or Dutch. They pre- sented eight tokens along a continuum from ‘‘da’’ to ‘‘ta’’ http://dx.doi.org/10.1016/j.cognition.2014.07.002 0010-0277/Ó 2014 Elsevier B.V. All rights reserved. ⇑ Corresponding author. Tel.: +61 2 9772 6493. E-mail address: paola.escudero@uws.edu.au (P. Escudero). Cognition 133 (2014) 408–413 Contents lists available at ScienceDirect Cognition journal homepage: www.elsevier.com/locate/COGNIT