An Investigation into Exemplar Effects in the Perceptual Learning Paradigm

Abstract

Variance in the speech signal is caused by many factors. Traditionally, it was assumed that variation is removed prior to lexical access. Evidence for abstraction has come from perceptual learning experiments. Exemplar Theory does not assume stored abstractions, but the storage of exemplars, as evidenced by same-speaker priming effects in which a listener is faster to react to a word when repeated in the same voice. This study combined the perceptual learning paradigm with a priming experiment to investigate whether exemplar effects are found on items which induce perceptual learning compared to canonically realised counterparts. After exposure to an ambiguous fricative in German between [f] and [s], categorisation changed as a function of condition, replicating previous studies. In the priming block, no difference in reaction-time was found between conditions, suggesting exemplars were not stored during perceptual learning. We discuss the results in relation to models of speech perception.