Perceptual processing of variable input in Spanish: An exemplar-based approach to speech perception

The effects of linguistic experience on the perceptual processing and identification of phonological dialect variation were investigated in a series of psycholinguistic experiments with native speakers of Spanish from Mexico and Puerto Rico. Perceptual processing of dialect variation was assessed using bisyllabic words produced by female speakers of Mexican and Puerto Rican Spanish with a speeded naming task and a lexical decision task. Identification of dialect variation was assessed using bisyllabic words with a two-alternative forced-choice classification task. The test stimuli used in all three tasks contained either a word-final /n/, a syllable-final /r/, or a syllable-final /s/. These phonological variables were chosen because they exhibit phonological variation to different degrees in the two dialects being studied here.; The results from the speeded naming task show a significant main effect for phonological variable, with words containing syllable-final /s/ resulting in the slowest naming (reaction) time. Factors that significantly interacted with other factors were sex, listener dialect, and speaker dialect. The results from the lexical decision task show a significant effect for phonological variable, where words containing syllable-final /s/ resulted again in the slowest reaction times. Interestingly, both Mexican and Puerto Rican participants were biased to label Mexican stimuli as a word, even when the stimuli were nonwords. This bias was not found for the Puerto Rican stimuli. The dialect identification task's results show that overall listeners most accurately identified the speaker's dialect when they produced words containing syllable-final /s/, while the dialect of speakers producing words containing syllable-final /r/ was least accurate. The Mexican listeners were more accurate at identifying their own dialect than they were identifying a dialect other than their own, as were the Puerto Rican listeners.; These results are easily modeled and accounted for within an exemplar-based approach. In an exemplar model, such as the one presented in Chapter 6, there are interactions and activations between several categories during speech perception and production. A listener's linguistic experience is stored as exemplars in the lexicon that are connected to other linguistic and extra-linguistic information that the listener has experienced. This way, input is stored as detailed exemplars, which activate, and in turn are activated by, other categories such as stereotypes (e.g. age, gender, dialect, etc.) and phonological generalizations (e.g. word-final nasals are velar). The findings of the current study add to the growing literature on the effects of linguistic experience on the perception of variable input, as well as to the growing literature on exemplar-based models of perception and production.