The voice of deceit: Comparing acoustic and perceptual data.

• Main Author: Rockwell, Patricia Ann.
• Other Authors: Buller, David
• Language: en
• Published: The University of Arizona. 1994
• Online Access: http://hdl.handle.net/10150/186929

This study examined the nature of deceptive vocal behavior in interactive situations. It compared those vocal features of deception that can be measured by acoustic equipment with those vocal features of deception that can be measured perceptually by human coders. As deception researchers traditionally measure vocal behavior with either acoustic or perceptual methods, it is uncertain what correspondence, if any, exists between these two methods. This study attempted to determine this correspondence. Deceptive interactions from an earlier study (Buller, Burgoon, Buslig & Roiger, 1993; Burgoon, Buller, Ebesu, White, and Rockwell, 1994) were used to conduct a detailed analysis of vocal features of deceptive speech. The vocal samples were analyzed perceptually and acoustically. Results indicated moderate correlations between some acoustic and perceptual variables, with neither measurement type proving conclusively superior to the other in discriminating between truth and deception. Of three categories examined (time, pitch, and intensity), the time variables of shorter message length, longer response latencies, slower tempo, and less fluency best discriminated between truthful and deceptive statements. Other variables that discriminated truth from deceit were increased intensity range, increased pitch level and variance, and less pleasant vocal quality. Analyses of deception type showed that fabricated deceptions were louder and lower pitched than equivocal deceptions. An analysis of deception planning, showed that planned deceptions exhibited more fluency, a lower pitch level, and less pitch variance than unplanned deceptions. An examination of correlations between deceiver/receiver evaluations of deceiver honesty and deceiver vocal behaviors showed moderate correlations occurred between these evaluations and length of response latencies, pitch level, pitch range, and pitch variance. In general, these findings provide further confirmation of Interpersonal Deception Theory.