"Auditory speech identification has been observed to be influenced by both lexical and visual information. Perceptual learning experiments have used two unique paradigms to test how each of these information sources affects the identification of ambiguous stimuli. In both cases, listeners are more likely to identify ambiguous stimuli in the direction of the disambiguating information they receive. It has been further argued that the resulting effects are the same and can be traced back to the same general speech perception mechanism. Despite this claim, there have been conflicting results in regards to generalization. Lexically induced perceptual learning has been observed to generalize to new contexts, while visually induced perceptual learning has been observed to be context dependent. While the difference in these observed results could be explained by the information source (lexical vs. visual), there are also crucial differences in the experimental designs that may offer a better account. The training stimuli set for lexically induced perceptual learning experiments includes many unique tokens that are presented one time each. For visually induced perceptual learning experiments, the training set includes just one unique token presented multiple times. Listeners therefore only receive type variation in the lexically induced perceptual learning experiments. Crucially, type variation has been observed to be necessary for learning linguistic patterns and therefore may explain the differences in observed results between the two paradigms. This current study uses three new experiments to study the generalizable nature of lexically induced perceptual learning. The results corroborate the idea that generalization of the effect to new contexts is possible in lexically induced perceptual learning experiments when listeners are trained with type variation, but when type variation is eliminated the ability to generalize the effect to new contexts is no longer observed."--Page ii.
Read
