Traditional methods of research on perceptual dialectology have been limited to tasks that ask participants to draw and label dialect regions on maps or to make attitude judgments about samples of certain linguistic varieties. Only a small handful of perceptual experiments have directly looked at how listeners identify and categorize where talkers are from based on actual speech samples. Our first experiment investigated how well naive listeners could categorize talkers based on regional dialect of American English using sentence-length utterances. Although listeners performed poorly on this task, a more detailed analysis of their error patterns revealed systematic perceptual confusions. The results suggest that listeners have knowledge of three broad dialect categories: New England, South, and North/West. Since listeners were able to perform the perceptual categorization task at levels reliably above chance, a second experiment was carried out to measure the acoustic-phonetic properties that listeners used to make their categorization judgments. Results of this acoustic analysis study revealed four robust acoustic-phonetic properties that were good predictors of where the talkers were actually from: New England r-lessness and /Q/ backing, North /ou/ offglide centralization, and South Midland /u/ fronting. These properties also appeared to be used by the listeners in making their perceptual judgments. A third experiment explored the effects of residential history on dialect categorization performance using two different groups of listeners with different linguistic experiences. Listeners who had lived in at least three different states comprised the “Army Brat” group; listeners who had lived only in Indiana comprised the “Homebodies” group. The results revealed that the Army Brats performed significantly better than the Homebodies on the perceptual dialect categorization task. This finding suggests that greater exposure to linguistic variation in development leads to better performance on the categorization task. A final experiment investigated the effects of short-term perceptual learning on categorization performance. Using the same talkers and response alternatives as in the first experiment, one group of listeners was trained to categorize one talker from each region and a second group was trained to categorize three talkers from each region. After training, both groups of listeners were then asked to categorize new talkers using the same six dialect regions. Results showed that listeners who were exposed to a range of variability after training with three talkers from each dialect region were better able to generalize to new talkers than the listeners who were trained to identify only a single talker from each region. Taken together, these four experiments provide new evidence for the important role of acoustic-phonetic variation and variability in speech perception and spoken language processing. The implications of these findings for speech perception and spoken language processing are discussed.
Stored only when both papers are in the corpus — 5 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.