Comprehension of acoustically degraded emotional prosody in Alzheimer's disease and primary progressive aphasia

Abstract

Previous research suggests that emotional prosody perception is impaired in neurodegenerative diseases like Alzheimer's disease (AD) and primary progressive aphasia (PPA). However, no previous research has investigated emotional prosody perception in these diseases under non-ideal listening conditions. We recruited 18 patients with AD, and 31 with PPA (nine logopenic (lvPPA); 11 nonfluent/agrammatic (nfvPPA) and 11 semantic (svPPA)), together with 24 healthy age-matched individuals. Participants listened to speech stimuli conveying three emotions in clear and noise-vocoded forms and had to identify the emotion being conveyed. We then conducted correlation analyses between task performance and measures of socio-emotional functioning. All patient groups showed significant impairments in identifying clear emotional prosody compared to healthy individuals. These deficits were exacerbated under noise-vocoded conditions, with all patient groups performing significantly worse than healthy individuals and patients with lvPPA performing significantly worse than those with svPPA. Significant correlations with social cognition measures were observed more consistently for noise-vocoded than clear emotional prosody comprehension. These findings open a window on a dimension of real-world emotional communication that has often been overlooked in dementia, with particular relevance to social cognition, and begin to suggest a novel candidate paradigm for investigating and quantifying this systematically.

Keywords: Alzheimer’s disease; Degraded speech; Dementia; Emotional prosody; Frontotemporal dementia; Noise-vocoding; Primary progressive aphasia; Social cognition.

Fig. 1

Boxplots of noise-vocoded emotional comprehension performance for each diagnostic group. Panel A shows data across participant groups; here, score refers to the percentage correct of the combined noise-vocoded prosody comprehension score, across all three vocoding channel levels. Significant between-group differences after adjusting for performance on identifying clear emotions and WASI matrix reasoning (as an index of disease severity) are coded with *** as p < 0.001, ** as p < 0.01 and * as p < 0.05. Panels B - F show performance profiles across vocoding channels (and in clear speech) for each participant group separately; score refers to the percentage correct of the prosody comprehension performance at each vocoding level. In all panels, the horizontal line within each box indicates the median score, with the boxes indicating the interquartile range; individual participant data points are superimposed. AD, patient group with typical Alzheimer’s disease; Control, healthy age-matched individuals; lvPPA, patient group with logopenic variant primary progressive aphasia; nfvPPA, patient group with nonfluent/agrammatic variant primary progressive aphasia; svPPA, patient group with semantic variant primary progressive aphasia.

Fig. 2

Correlation plots of clear emotional prosody comprehension with measures of social cognition across the patient cohort. This Figure shows how different standard measures of social cognition were correlated with total score on recognition of clear (natural) emotional prosody (see text) across syndromic groups, as follows: (A) correlation with the full modified Interpersonal Reactivity Index (mIRI); (B) correlation with the cognitive empathy subscale in mIRI; (C) correlation with the cognitive empathy (perspective taking) subscale in mIRI; (D) correlation with the full revised self-monitoring scale (RSMS); (E) correlation with the sensitivity to socio-emotional expressiveness RSMS subscale; (F) correlation with the monitoring self-presentation RSMS subscale. Spearman’s rank and p-value shown alongside each correlation line, bold green font indicates a significant correlation (p < 0.05). The percentage correct here is the percentage correct for clear emotional prosody (control task). Dots represent individual participants’ performance, with colours representing each syndromic diagnosis, as coded in the key (right); shading represents 95% confidence intervals. AD, Alzheimer’s disease; lvPPA, logopenic variant primary progressive aphasia; mIRI, modified Interpersonal Reactivity Index; nfvPPA, nonfluent variant primary progressive aphasia; RSMS, revised self-monitoring scale; svPPA, semantic variant primary progressive aphasia.

Fig. 3

Correlation plots of noise-vocoded emotional prosody comprehension with measures of social cognition across the patient cohort. This Figure shows how different standard measures of social cognition were correlated with total score on recognition of noise-vocoded emotional prosody (see text) across syndromic groups, as follows: (A) correlation with the full modified Interpersonal Reactivity Index (mIRI); (B) correlation with the cognitive empathy subscale in mIRI; (C) correlation with the cognitive empathy (perspective taking) subscale in mIRI; (D) correlation with the full revised self-monitoring scale (RSMS); (E) correlation with the sensitivity to socio-emotional expressiveness RSMS subscale; (F) correlation with the monitoring self-presentation RSMS subscale. Spearman’s rank and p-value shown alongside each correlation line, bold green font indicates a significant correlation (p < 0.05). The percentage correct here is the combined noise-vocoded score (combined across all three levels of vocoding channels: see text). Dots represent indvidual participants’ performance, with different colours representing each syndromic diagnosis, as coded in the key (right); shading represents 95% confidence intervals. AD, Alzheimer’s disease; lvPPA, logopenic variant primary progressive aphasia; mIRI, modified Interpersonal Reactivity Index; nfvPPA, nonfluent variant primary progressive aphasia; RSMS, revised self-monitoring scale; svPPA, semantic variant primary progressive aphasia.