Abstract # 4505 Event # 100:

Scheduled for Friday, June 21, 2013 10:00 AM-10:15 AM: Session 11 (San Geronimo Ballroom B) Oral Presentation


L. A. Heimbauer1,2, M. J. Beran2 and M. J. Owren3
1Pennsylvania State University, Department of Psychology, 442 Moore Building, University Park, PA 16802-3106, USA, 2Language Research Center, Georgia State University, 3Emory University
     We assessed a language-trained chimpanzee’s ability to recognize time-reversed speech—a synthetic form whereby segments of equal millisecond (ms) length, called “windows,” are reversed in time. The manipulation preserves amplitude of each frequency component, but reverses this pattern of energy within each window. Human listeners are able to recognize time-reversed speech at window lengths up to 100 ms; however, disruption of the amplitude envelope at longer lengths reduces intelligibility (Saberi & Perrott, 1999). The theoretical interpretation is that individual phonetic segments of natural speech typically range from 50 to 100 ms (Crystal & House, 1988), and reversal-windows less than 100-ms leave phoneme perception undisturbed. We hypothesized that because Panzee identifies speech in both natural and other synthetic impoverished forms (Heimbauer, Beran, & Owren, 2011), she perceives speech based on phonemic segments, as humans do. Panzee and human participants (Homo sapiens) were tested with twenty words, each synthesized in eight forms, using reversals ranging from 25 ms to 200 ms. Results revealed that for both species time-reversal window length significantly predicted percentage-correct word identification, with 50% intelligibility at approximately 130-ms window length. We concluded that Panzee is sensitive to phonemic cues in time-reversed, and, hence, natural speech. For this animal, like humans, detailed auditory analysis of the short-term acoustic spectrum is not essential. Rather, the amplitude envelope and phoneme-length information are more likely the important factors.