Disentangling Frequency and Talker Variability in a Statistical Learning Task Investigating Child Speech Acquisition

Abstract

Past research suggests that statistical patterns in a child’s native language influence the child’s speech production accuracy (Stoel-Gammon, 1998; Plante, Bahl, Vance, & Gerken, 2011). In addition to hearing some forms more often than others, children also hear forms from a variety of talkers in their environments (parents, caretakers, etc.). This variable, known as talker variability, can have facilitative effects on children’s ability to reproduce nonwords (Plante et al., 2011; Richtsmeier, Gerken, Goffman, and Hogan, 2009).To further investigate frequency and talker variability, the current study employed a within-subjects design to expose 3-4 year olds to four levels of experimental frequency, with and without talker variability. The results of this study suggest a benefit for perceptual frequencies greater than 1, and for production practice. Benefits from talker variability should not be ruled out, but may be less robust than basic perceptual frequency.