Shaping perceptual learning of synthetic speech through feedback
- PDF / 473,896 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 113 Downloads / 235 Views
Shaping perceptual learning of synthetic speech through feedback Matthew I. Lehet 1 & Kimberly M. Fenn 1 & Howard C. Nusbaum 2,3
# The Psychonomic Society, Inc. 2020
Abstract Listeners exposed to accented speech must adjust how they map between acoustic features and lexical representations such as phonetic categories. A robust form of this adaptive perceptual learning is learning to perceive synthetic speech where the connections between acoustic features and phonetic categories must be updated. Both implicit learning through mere exposure and explicit learning through directed feedback have previously been shown to produce this type of adaptive learning. The present study crosses implicit exposure and explicit feedback with the presence or absence of a written identification task. We show that simple exposure produces some learning, but explicit feedback produces substantially stronger learning, whereas requiring written identification did not measurably affect learning. These results suggest that explicit feedback guides learning of new mappings between acoustic patterns and known phonetic categories. We discuss mechanisms that may support learning via implicit exposure. Keywords Language comprehension . Implicit vs. explicit memory . Perceptual learning . Psycholinguistics
Introduction Native language acquisition is not thought of as explicitly instructed, yet children learn mappings between acoustic features in the speech signal and lexical representations (Idemaru & Holt, 2013; Nittrouer, Lowenstein, & Packer, 2009). There is longstanding debate over whether these mappings are acquired through simple exposure, suggesting an innate propensity to acquire spoken language (e.g., Chomsky, 1965; Pinker, 1994), or whether some feedback is necessarily part of acquisition (e.g., Demetras, Post, & Snow, 1986). Various studies suggest that adults use information in their speech environment to adjust mappings between acoustics and lexical representations, allowing dynamic on-line adjustments to speech categories (Guediche, Fiez, & Holt, 2016; Idemaru & Holt, 2014; Kleinschmidt & Jaeger, 2015; Kraljic & Samuel, 2006; Vroomen, van Linden, de Gelder, & Bertelson, 2007).
* Matthew I. Lehet [email protected] 1
Department of Psychology, Michigan State Universit, East Lansing, MI, USA
2
Department of Psychology, The University of Chicago, Chicago, IL, USA
3
Center for Cognitive and Social Neuroscience, The University of Chicago, Chicago, IL, USA
However, the ways in which internally generated predictions and externally provided feedback influence perceptual learning remain unclear. Across a variety of paradigms, environmental information helps listeners adjust mappings between acoustic features and lexical representations. This learning can happen explicitly through orthographic feedback that informs the mapping between distorted acoustics and lexical representations (Davis, Johnsrude, Hervais-Adelman, Taylor, & McGettigan, 2005; Fenn, Margoliash, & Nusbaum, 2013; Guediche et al., 2016; Hervais-Adelman, Davis, Johnsrude
Data Loading...
