The Older Voice: Traits and Their Acoustic Characteristics

EMERSON COLLEGE
Bringing Innovation to Communication and the Arts
Daniel Kempler1 
The Older Voice: Traits
and Their Acoustic Characteristics
Tracy McLaughlin-­‐Volpe1  Joann Montepare2
1Emerson College With Kimberly Oaks, Emily O’Sullivan, & Michelle Steingiser1 2Lasell College Abstract Speakers are judged by the sound of their voices. This study compared how younger vs. older speakers are perceived in terms of personality, interacEon intenEons and health. Older speakers were judged to be less healthy and less engaged but wiser than younger speakers. AcousEc cues underlying these judgments are explored. Trait Ratings & Acoustic Correlates
Perceived Age Purpose The present research was designed to examine how voices elicit impressions of aging men and women. We were parEcularly interested in the correlaEons of acousEc variables with both posiEve and negaEve impressions of individuals based on their voices. Speakers 80 Age Es*mate Age stereotypes abound, and age-­‐related physical characterisEcs to contribute significantly to percepEons of aging adults (Nelson, 2002; Montepare & Zebrowitz, 1998). In addiEon, it is well-­‐
known that voices change with age in relaEvely consistent and predictable ways (e.g., Harnsberger et al., 2008, Linville, 2001). 70 60 50 40 30 20 10 0 20 30 40 50 60 70 80 Speaker Age •  Listeners esEmated the age of speakers accurately (r=.90, p<.000). •  Judgments of age-­‐related percepEons (e.g. “elderly-­‐youthful”) were highly correlated with chronological age and age-­‐esEmates (p <.001) •  The prominent acousEc cues associated with older age are low pitch, high amplitude perturbaEon and slow speech rate. Perturba*on
AGE
Age
Ji#er (sPPQ)
Pitch
Rate
**
-­‐.22
.22
.07
.03
**
-­‐.34*
.27
.13
.11
.64**
.31
-­‐.21
-­‐.06
-­‐.05
-­‐.65**
.27
.38*
-­‐.72
Direct age esJmate
Youthful
.90**
.27
.43*
-­‐.91**
-­‐.36*
-­‐.51**
-­‐.68
**
.67
PHYSICAL CHARACTERISTICS Age
Ji#er (sPPQ)
-­‐.63**
-­‐.21
Chronological age
Intensity
Energy Shimmer Mean F0 F0 Variability Mean SEmulus Sentence Variability (sAPQ) (z-­‐score, Hz) (SD, Hz) Energy (dB)
Length (sec) Length (sec)
(SD, dB)
.66**
Method   13 male and 13 female volunteers, 22 – 79 years old   Reading a script: “I'd like you to listen to my voice. AVer you hear it, answer the quesJons about what I sound like. 1 2 3 4 5 6 7 8 9 10”. Listeners   96 students (23 men and 73 women) parEcipated as raters in exchange for course credit. Physical Voice CharacterisJcs 7 Healthy Voice Scale   NaEve-­‐English speakers 6 6 5 5 4 4 3 3 2 20 30 Procedures   Listeners were told that the purpose of the study was to invesEgate the impressions people form based solely on listening to voices and that they would rate a series of 13 male (or female) voices on different scales.   The voices were presented one at a Eme with a 5 second pause separaEng the voices. Listeners rated each voice on scales reflecEng personality traits, interacEon intenEons, physical vocal characterisEcs and perceived age. Circle the number that best describes the speaker you just heard. We’re interested in your gut reacJons. Dull 1 2 3 4 5 6 7 Dynamic 50 60 70 80 7 Personality Traits 6 6 Wisdom Examples of raJng scales: 40 Speaker Age 5 5 4 4 3 3 2 20 30 40 50 60 70 80 Speaker Age Circle the number that best describes how you would feel about interacJng with the speaker. “I would avoid this person.” Strongly Agree 1 2 3 4 5 6 7 Strongly Disagree Analyses   CorrelaEons were computed between independent variables (age and acousEc data) and the dependent variables (personality traits, interacEon intenEon and voice health raEngs)   The sEmuli and the mid-­‐porEon (2sec) of sustained vowel samples provided by the speakers were analyzed using the Kay MulEspeech and MulEdimensional Voice Program (MDVP) to yield measurements for (1) pitch perturbaEon (sPPQ), (2) intensity perturbaEon (sAPQ), (3) mean pitch*, (4) mean intensity, (5) variability of pitch (6) variability of intensity and (7) speech rate/sEmulus length. (*Z-­‐scores were computed for men and women separately for average pitch measures) •  Older speakers were judged to be less engaged and more rigid, but wiser than younger speakers. •  VariaEon in loudness appears to be the strongest correlate of overall engagement, while slower speech correlates more strongly with higher wisdom raEngs. Perturba*on
Healthy Voice (smooth/
clear/stable)
High pitch
-­‐.69**
A#racEve
-­‐.36*
•  Young listeners were less inclined to interact with older than younger speakers. •  Listeners were more likely to want to interact with people who used more prosody and spoke slower overall. 6 6 5 5 4 4 3 3 2 20 30 40 50 60 Speaker Age 70 80 Pitch
Engaged (engaged, dynamic, ac*ve, happy) Warmth (warm, caring, kind) Wisdom Age **
-­‐.37*
.52
-­‐.15
-­‐.29
.79
**
.16
-­‐.05
.52
**
Ji#er ( sPPQ) INTENTIONS
Approach Scale Comfort Scale
Good story teller
Rate
.56**
-­‐.13
.22
.11
-­‐.36
.41*
-­‐.24
.02
.07
-­‐.33
.23
.01
.20
.24
-­‐.28
Pitch Intensity Rate Energy SEmulus Sentence Shimmer Mean F0 F0 Variability Mean Variability Length (sec) Length (sec) (sAPQ) (z-­‐score, Hz) (SD, Hz) Energy (dB) (SD, dB) **
-­‐.48** .12 -­‐.25 .51 .51** .01 .42* .11 -­‐.10 .29 -­‐.14 .22 .32 -­‐.01 .58** .43* 0.19 .76** .36* .30 -­‐.63** .04 .28 .37* .27 0.72** Perturba*on
INTERACTION Intensity
Energy Shimmer Mean F0 F0 Variability Mean SEmulus Sentence Variability (sAPQ) (z-­‐score, Hz) (SD, Hz) Energy (dB)
Length (sec) Length (sec)
(SD, dB)
Perturba*on PERSONALITY TRAITS InteracJon IntenJons 7 Approach Scale   Intraclass correlaEon coefficients (ICC) computed for each scale item, separately for the set of female and male voices revealed consensus among parEcipants in raEngs of the individual voices. •  Older speakers were judged to be less healthy than younger speakers on raEng of vocal health (e.g.,“rough-­‐smooth”, “raspy-­‐clear”, “sick-­‐healthy” and “unstable-­‐stable.”) •  Higher vocal health raEngs were associated with with lower pitch and higher pitch perturbaEon. Age
-­‐.47**
-­‐.46**
.11
Ji#er (sPPQ)
.15
.03
.27
Pitch
Intensity
-­‐0.32 Rate
Energy Shimmer Mean F0 F0 Variability Mean SEmulus Sentence Variability (sAPQ) (z-­‐score, Hz) (SD, Hz) Energy (dB)
Length (sec) Length (sec)
(SD, dB)
-­‐.22
-­‐.24
-­‐.17
.29
.26
-­‐.09
.30
.24
.23
-­‐.14
-­‐.14
.09
.46**
.46**
.62**
.51**
.52**
.59**
-­‐0.14
-­‐0.06
0.26