Tipultech logo

Recognition of Prosody

Dawn Senathi-Raja

Given the prima facie similarities between prosody and music, a task developed by Patel and colleagues (Patel et al., 1998) was administered to examine MR's perception of prosody for signs of deficits analogous to those he exhibits for music. The stimuli consisted of sentence pairs that differed in prosody, as well as analogous music pairs. As shown in Appendix J, there were three types of sentence pairs: (1) Statement-question pairs differed in terminal-pitch information (i.e. questions had a rising pitch, while statements had a falling pitch); (2) Emphasis-shift pairs differed in internal-pitch information (i.e. different words in a sentence varied in pitch depending on whether they carried the emphasis); and (3) Timing- shift pairs differed in the placement of pauses. Musical analogues of statement-question pairs

Table 2.6
Percentage of Errors in Beckwith's (2003)Environmental Sounds Identification Task

Sound category MR's errors (Beckwith, 2003) MR's errors in the present study Controls errorsb Mean (SD)
Human non-verbal (e.g. baby crying) 66.67%a 50.00% 26.76% (13.3)
Man-made (e.g. sawing wood) 0.00% 0.00% 0.00% (0.0)
Animals (e.g. dog barking) 0.00% 0.00% 0.00% (0.0)
Technical (e.g. helicopter) 35.71%*** 35.71%*** 8.00% (7.5)
Naturally occurring (e.g. wind) 75.00%***a 25.00%*** 0.00% (0.0)
Total 32.26%*** 22.58%*** 6.53% (4.6)

a MR's illusions prevented correct responses for a number of stimuli in this category.
b MR's five matched controls completed this task.
***p < 0.001

differed in terminal melodic contour. Emphasis-shift musical analogue pairs varied in internal melodic contour. Finally, musical analogues of Timing-shift pairs were rhythmically different. Each of the six tasks contained 32 trials.

Prosodic and Musical Discrimination Tasks

Linguistic and musical discrimination tasks required participants to decide whether the stimuli within a pair were the "same" or "different". MR performed within the normal range for all linguistic tasks, indicating preserved speech prosody representation. His performance for musical discrimination stimuli was not significantly different from the comparison group except for the Emphasis-shift musical pairs (see Table 2.7). This indicates that MR was unable to use internal contour cues to facilitate his performance, which is consistent with his impaired performance on melodic contour tasks based on internal contour differences. Although MR's performance fell within the normal range for the musical Timing-shift condition, a chi-square goodness-of-fit one-way test revealed that his score was not significantly better than chance (?2 ? 2.00, ns), indicating that he was unable to reliably discriminate between rhythmic stimuli. This result is consistent with his impaired rhythmic ability on musical tasks (Beckwith, 2003). Examination of MR's scores across the Statement- question and Emphasis-shift melodic tasks highlighted a disproportionately large amount of errors (66.6%) for identical melodic pairs, which MR perceived as different. This finding was also apparent for MR's musical assessment on Trehub et al.'s (1984) identical melodic contour pairs, pointing to a low level perceptual problem.

Table 2.7

Percentage of Correct Responses for Patel et al.'s (1998) Prosodic and Musical Discrimination Tasks

Statement-question pairs Emphasis-shift pairs Timing-shift pairs

Speech Music Speech Music Speech Music
MR 100.00% 93.75% 100.00% 50.00%* 93.75% 62.50%

Mean 94.81% 93.50% 96.12% 84.11% 91.50% 80.25%
SD 5.12% 5.67% 5.76% 13.50% 10.33% 10.70%

a Data based on 12 controls, matched for age to MR (taken from Nicholson et al., 2003)
*p < 0.05

Prosodic Comprehension Tasks

Statement-question stimuli required that participants use prosodic information to categorise a sentence as a statement or question in the absence of other lexical cues. The Emphasis-shift task tested participants' abilities to use pitch information to detect which word in a sentence carried the emphasis. Finally, the Timing-shift task required participants to use prosodic information to interpret syntactically ambiguous questions (refer to Appendix K for details). Each task was comprised of 32 trials. As shown in Table 2.8, MR's performance was within the normal range for all prosodic perception tasks indicating preserved speech intelligibility relevant to the prosody domain.

Table 2.8

Percentage of Correct Responses for Patel et al.'s (1998) Prosodic Comprehension Tasks

Statement-question Emphasis detection Timing shift
MR 96.88% 100.00% 90.63%

Mean 99.00% 98.25% 86.72%
SD 2.67% 3.75% 10.40%

<< >>