Experimental support for a one-step model of phoneme acquisition

Abstract

Phonetic category acquisition involves a distributional learning mechanism (Maye, Werker, and Gerken 2002). Some researchers suggest that phonetic category acquisition is only the first step in a two-step model of phonological acquisition by modelling these two steps separately (Guenther and Gjaja, 1996; Boersma, Escudero, and Hayes, 2003; Peperkamp, Pettinato, and Dupoux 2003; Peperkamp, Calvez, Nadal, and Dupoux 2006), while others have argued for a one-step model (Dillon, Dunbar, and Idsardi 2013). This experimental study maps the learning trajectory of three groups of adult learners: (1) a group exposed to a bimodal frequency distribution where both halves of the bimodal distribution occur in complementary environments (Bimodal-Comp group), (2) a group exposed to a bimodal frequency distribution where both halves of the bimodal distribution occur in non-complementary environments (Bimodal-NonComp group), and (3) a group exposed to a monomodal frequency distribution (Monomodal group). This study finds support for a one-step model of phoneme acquisition, with the Bimodal-Comp group having lower sensitivities to critical stimuli than even the Monomodal group at all three exposure times tested.