Eliciting Speech Reduction Ii

This experiment extended work done previously in our laboratory to develop a method to elicit from talkers three different speaking styles, reduced, citation, and hyperarticulated, using controlled materials in a laboratory setting. In the initial experiment, the reduced style was elicited by having subjects read a sentence while carrying out a distractor task that involved recalling a fixed number of digits from shortterm memory. The original experiment was clearly limited in its success at eliciting a reduced style of speech: Only one of the six talkers showed significant differences between reduced and citation speech based on an acoustic analysis of the sentences. In this study, we chose to calibrate the distractor memory task to an individual's short-term memory span as measured by a simple digit span task. That is, the number of digits to be recalled after reading aloud a test sentence was determined by each individual's digit span. Twelve talkers were recorded in this experiment. The results showed that six of the twelve talkers produced a reduced style of speech for the test sentences in the distractor task relative to the same sentences in the citation style condition, as determined by a phonetically-trained judge. This initial evaluation was confirmed in a perceptual test using a pairwise comparison task in which normal-hearing, untrained listeners were presented with two sentences varying in speech style and were asked to choose the most carefully pronounced sentence. The results showed that 71% 88% of the sentence sets tested were correctly differentiated by speech style, indicating that the individual calibration method was a substantial improvement over the elicitation method of the original experiment.