Date of Award

5-2001

Level of Access

Open-Access Thesis

Degree Name

Master of Arts (MA)

Department

Communication Sciences and Disorders

Advisor

Judy P. Walker

Second Committee Member

D. Kimbrough Oller

Third Committee Member

Marisue Pickering

Abstract

One mild AOS, one moderate AOS and one control speaker were asked to produce utterances with different emotional intent. In Experiment 1, the three subjects were asked to produce sentences with a happy, sad, or neutral intent through a repetition task. In Experiment 2, the three subjects were asked to produce sentences with either a happy or sad intent through a picture elicitation task. Paired t-tests comparing data from the acoustic analyses of each subject's utterances revealed significant differences between FO, duration, and intensity characteristics between the happy and sad sentences of the control speaker. There were no significant differences in the acoustic characteristics of the productions of the AOS speakers suggesting that the AOS subjects were unable to volitionally produce acoustic parameters that help convey emotion. Two more experiments were designed to determine if näive listeners could hear the acoustic cues to signal emotion in all three speakers. In Experiment 3, näive listeners were asked to identify the sentences produced in Experiment 1 as happy, sad, or neutral. In Experiment 4, näive listeners were asked to identify the sentences produced in Experiment 2 as either happy or sad. Chi-square findings revealed that the naive listeners were able to identify the emotional differences of the control speaker and the correct identification was not by chance. The näive listeners could not distinguish between the emotional utterances of the mild or moderate AOS speakers. Higher percentages of correct identification in certain sentences over others were artifacts attributed to either chance (the näive listeners were guessing) or a response strategy (when in doubt, the naive listeners chose neutral or sad). The findings from Exp. 3 & 4 corroborate the acoustic findings from Exp. 1 & 2. In addition to the 4 structured experiments, spontaneous samples of happy, sad, and neutral utterances were collected and compared to those sentences produced in Experiments 1 & 2. Comparisons between the elicited and spontaneous sentences indicated that the moderate AOS subject was able to produce variations of FO and duration similar to those variations that would be produced by normal speakers conveying emotion (Banse & Scherer, 1996; Lieberman & Michaels, 1962; Scherer, 1988). The mild AOS subject was unable to produce prosodic differences between happy and sad emotion. This study found that although these AOS subjects were unable to produce acoustic parameters during elicited speech that signal emotion, they were able to produce some more variation in the acoustic properties of FO and duration, especially in the moderate AOS speaker. However, any meaningful variation pattern that would convey emotion (such as seen in the control subject) were not found. These findings suggest that the AOS subjects probably convey emotion non-verbally (e.g., facial expression, muscle tension, body language).

Share