Comprehension of Emotional Prosody in Parkinson’s Disease


Kelly Davis Garrett,+* Christine Lee,* Corey McMillan,* Ariel Goldberg,* Douglas L. Chute,+
Mark Liberman^ & Murray Grossman*^


+Neuropsychology, Drexel University

*Dept. of Neurology, University of Pennsylvania

^Dept. of Linguistics, University of Pennsylvania




While the dysprosody associated with Parkinson’s disease (PD) has been well described, findings from investigations on the comprehension of emotional intonation have been mixed. In the present investigation, we sought to examine the comprehension of emotional and non-emotional prosody among individuals with PD (n=14), neurologically healthy age and education-matched elders (n=17), and college-age adults (n=16). Stimuli included audio recordings of professional actors rendering utterances of minimal semantic content (i.e., dates, numbers) using emotional attitudes (tentative to dominant) and non-emotional attitudes (speaker close to intended listener, at a conversational distance, and far apart). Participants were instructed to rate 90 audio stimuli of emotional attitudes and an additional 90 stimuli of distance stimuli using 5-point Likert scales. Between-group comparisons showed that PD patients and control subjects do not differ in their judgments of distance stimuli, but PD patients and non-demented older adults are less accurate than younger control subjects in their judgments of emotional prosody (f[2,42]=5.6,p=.007). Within-group comparisons revealed poorer performance for emotional prosody judgments than non-emotional prosody judgments in the PD group (t[13]=-4.9,p<.001) and older adults (t[15]=-2.4; p=.008); the young control group did not differ in judgments of emotional and non-emotional prosody. There was no difference between groups in judgments of fundamental perceptual elements of prosody (pitch, duration, volume). Statistically significant correlations between emotional prosody comprehension and working memory were observed for the PD group (r = -.42, p = .003), but not for the control groups. These findings support the hypothesis that degradation of the frontal-striatal circuit in PD compromised the interpretation of emotional prosody comprehension. Factors contributing to this deficit include limited executive resources and poor comprehension of multi-modal emotional stimuli.

Garrett, K.D., Koenig, P., Moore, P., Devita, C., Listerud, J., Liberman, M. & Grossman, M. (February, 2002). Neurophysiological correlates in the comprehension of emotional prosody: An ƒMRI study. Journal of the International Neuropsychology Society, 8(2), 139.


Prosody- the variations in melody, intonation, pauses, stresses and accents of speech- can communicate both linguistic intent and emotional state.

Initial clinical and experimental reports argued for a right hemisphere specialization for prosody comprehension. However, more recent investigators have disputed this, arguing specialization for acoustic elements of prosody (i.e., rhythm, pitch) rather than a paralinguistic analogue to left cortical language areas.

Some investigators of the neuropsychology of prosody have argued that dissociations between emotional and syntactic prosody among neurologic patients may- in part- be due to differences in task demands, rather than specialization of cortical areas for interpretation of a speakers emotional state and linguistic intent. Furthermore, interpreting emotional prosody may involve more sophisticated integration of multiple elements of the acoustic signal (e.g., pitch, volume, duration) than non-affective prosody.

The comprehension of prosody among Parkinson’s disease patients (PD) without dementia has not been well documented. PD provides an interesting model of the comprehension of prosody because one may investigate the role of executive functions, especially working memory, and frontal-striatal functioning on interpretation of emotions and emotional prosody.

If deficits in emotional comprehension are demonstrated among individuals with PD, it is important to explore the relative contributions of:

  • primary perceptual skills
    (e.g., ability to appreciate differences in rate, pitch contours, volume)
  • executive functions required to interpret emotions
  • affective competence, or the ability to feel, express and understand emotions from within and to accurately interpret emotional states of others based on their behaviors (e.g., verbal communication, prosody, affect, gestures, facial expressions).


Demographic descriptions of early-stage Parkinson’s without dementia, non-demented older adults, and young adults. (MMSE = Mini-mental state exam; BDI-2 = Beck Depression Inventory, 2nd edition). Mean (Standard Deviation).


 Parkinson’s (n=14)

 Older Adults (n=17)
 Young Adults (n=16)


 66.4 (9.3)
 69.6 (8.2)
 20.6 (2.2)


 15.4 (2.8)
 15.8 (2.6)
 14.6 (1.6)

 % Male



 29.0 (0.9)
 29.1 (1.1)
 30 (0)


 8.1 (5.7)
 3.9 (3.9)
 6.2 (4.7)


Materials: Background Measures


Facial Recognition Test- Penn Revision (Glosser et al., under review)-
A 20-item, two choice match-to-sample task (stimuli from Benton et al., 1978).








Penn Penn Emotional Recognition Test (Gur et al., 2001)-
A 40-item, forced choice measure of emotional face identification.


What emotion is this face showing?

Click on the word to the right that best describes the emotion.





No Emotion

Letter-Number Sequencing (WMS-III; Weschler, 1997)-
A measure of auditory working memory.


Materials: Experimental Measures

Penn Continuum of Emotional Prosody- Stimuli consisted of audio recordings of speech samples from 4 professional actors (2 males, 2 females) for three tasks: Acoustic Discrimination, Emotional Prosody, and Non-emotional Prosody. Dependent measures for each were the sum of absolute differences between an individual’s judgment and normative ratings for each item.

  • Acoustic Discrimination- Participants listened to repetitions of a syllable ("ma-ma ma-ma") and judged whether the last utterance is the same, more or less intense than the rest of the utterance using a 5-point Likert scale. Stimuli were manipulated for separate runs of duration, pitch, and volume.
  • Emotional Prosody- Participants listened to semantically neutral, 4-syllable gain-equalized utterances (dates and numbers) and judged the emotional tone using a 5-point Likert scale. This scale was inspired by the comparative literature on prosody and the poor psychometric properties of category based emotional rating paradigms among neurologically healthy adults.














  • Non-emotional Prosody&emdash; Participants listened to semantically neutral, 4-syllable gain equalized utterances (dates and numbers) and judged the interlocutor distance between speaker and intended listener using a 5-point Likert scale. This paradigm was created to evaluate the ability to appreciate non-emotional, non-linguistic manipulations of the acoustic elements of prosody.



Tête á tête or

Under breath

Conversational Distance

Distant or

with ambient noise


Parkinson’s patients were not different from young or older adults in their ability to discriminate changes in volume, duration, and pitch of speech sounds or faces on a version of the Benton Facial Recognition Test. However, Parkinson’s patients (t[25] = 2.6, p=.02) and older adults (t[27] = 3.2, p=.003) were less accurate in the identification of emotional faces than younger adults.


Summary of performance on background measures: Mean (Standard Deviation).


 Older Adults
 Young Adults

 Benton Facial Recognition

 20.8 (2.5)
 20.8 (3.4)
 21.7 (1.6)

 Penn Emotional Recog. (Faces)

 30.1 (3.7)
 30.7 (3.3)
 34.1 (2.2)

 Letter-Number Sequencing

 10.9 (3.1)
 10.4 (3.2)
 13.4 (2.0)





Originally presented at the International Neuropsychology Society Meeting, February 2002, Toronto, ON, CANADA.

Return to Kelly Davis Garrett's Home Page