- Мандрилы научились «вынюхивать» больных сородичей [2017-05-15]
- Хищные муравьи оказались отличными спасателями [2017-05-15]
- Биологи впервые записали все «мысли» гидры [2017-05-15]
- Стресс заставляет людей сопереживать другим [2017-05-15]
- Передача знаний по наследству улучшила навигацию почтовых голубей [2017-05-15]
- В компании раки «напиваются» в полтора раза быстрее [2017-05-15]
- Когда пригорает: «социологизаторы» против «биологизаторов» [2016-06-13]
- «Женские секреты приматов» от Евгении Тимоновой. Насколько достоверна эта популярная передача? [2016-06-06]
Large sex differences characterize the human voice and its anatomical substrates. Men's fundamental frequency (F0), the primary acoustic correlate of pitch, is approximately half as high as women's (Titze, 2000). F0 is inversely related to vocal fold length, which is 60% longer in men (Titze, 2000), far greater than the roughly 7% difference in height (Gaulin & Boster, 1985). Formant frequencies, which affect perceptions of voice timbre (Cleveland, 1977), are more closely spaced in men (Rendall, Kollias, Ney, & Lloyd, 2005). This spacing, called formant dispersion (Df), is inversely related to vocal tract length (Fitch & Hauser, 1995), which is greater in men, both absolutely and relative to height (Fitch & Giedd, 1999).
Vocal sex differences emerge at puberty when high testosterone levels cause males' vocal folds and tracts to grow faster than overall body growth (Fitch & Giedd, 1999, Lee et al., 1999). No comparable changes occur in females. The evolutionary reasons for these sex differences are unclear, but several lines of evidence indicate that sexual selection specifically shaped men's voices (Collins, 2000, Feinberg et al., 2006, Feinberg et al., 2005, Puts, 2005, Puts et al., 2006).
Sexual selection can operate through mate choice. Women prefer deeper male voices (Collins, 2000, Feinberg et al., 2005, Puts, 2005), especially near ovulation (Feinberg et al., 2006, Puts, 2005) and when evaluating short-term mates (Puts, 2005). Male–male dominance competition may also have played a role (Puts, Gaulin, & Verdolini, 2006). Manipulations of both F0 and Df independently affect women's ratings of speakers' age, masculinity, and size (Feinberg et al., 2005), and these parameters might also affect perceptions of dominance among men. Indeed, men rate male voices lowered in both F0 and Df as more dominant than the same voices with these acoustic parameters raised (Puts, Gaulin, & Verdolini, 2006). These manipulations affect ratings of physical dominance (fighting ability) more than ratings of social dominance (respect among peers, leadership, etc.). However, because both F0 and Df were manipulated simultaneously by Puts, Gaulin, Verdolini, & Hodges, (2006), it is impossible to determine whether F0, Df, or both, affected perceptions of dominance among men—an issue central to understanding how sexual selection shaped men's voices. Thus, we examine the relative contributions of F0 and Df to ratings of physical and social dominance by manipulating F0 and Df independently and by similar perceptual amounts.
Fifty male and 20 female (ages 18–24) English-speaking, normally hearing undergraduates at UC Santa Barbara participated in one of three IRB-approved studies. Two were preliminary just-noticeable-difference (JND) studies; the third was the focal study.
2.2. Studies 1 and 2: JND studies
Experiments were computerized (E-Prime), and voice parameters were manipulated using Praat 4.4.06.
A common measure of perceptual magnitude was required to compare the relative effects of F0 and Df on dominance ratings. Using Fechnerian and Weberian psychophysical principles (Marks & Gescheider, 2002), we chose the JND. Although previous studies have examined the JND for various acoustic parameters (e.g., Smith, Patterson, Turner, Kawahara, & Irino, 2005), results vary due to methodological differences. The JND will depend on stimulus magnitude as well as details of instrumentation. Thus, we obtained our own JNDs—tailored to our equipment, methods, and the magnitude of the stimuli used in Study 3—for both F0 and Df.
Twenty-eight subjects (8 men, 20 women) participated in a JND study of either F0 or Df. For both F0 and Df studies, 30 pairs of voice stimuli were created, comprising an equal mix of 10 specific manipulations, including a null manipulation. F0 manipulations ranged from 0.6 to 2.2 semitones in nine increments of 0.2 semitones, plus unmanipulated recordings. Df manipulations ranged from 2.5% to 6.5% changes in apparent vocal tract length in nine increments of 0.5%, plus unmanipulated recordings. For F0 manipulation, voices were raised and lowered from baseline levels without affecting tempo. For Df manipulation, formant structure was shifted up and down (increasing or decreasing Df) without affecting tempo. Where F0 was affected by this procedure, it was manipulated back, using the procedures described above, to within 0.2 semitones (<0.17 JND) of its original, unmanipulated values. For both F0 and Df manipulation, parameters were set to a time step of 0.01, a minimum pitch of 75 Hz, a maximum pitch of 300 Hz, and otherwise to default.
Participants were directed through an experimental session where paired voice stimuli were presented in random order. Each stimulus pair consisted of (A) a 3- to 5-s voice clip of an 18- to 24-year-old man introducing himself, followed by (B) the same clip at the same or different F0 or Df. AB pairs were presented three times before a response was permitted. Participants were instructed to determine if Clip B was the same as or different from Clip A. We defined the JND as the smallest increment in F0 or Df for which 50% of subjects perceived a difference. For F0, this was 1.2 semitones; for Df, it was a 4% change.
2.3. Study 3: voice effects on dominance
Information on JNDs was used to manipulate 30 voice recordings selected randomly from 111 produced in a previous “dating-game” experiment (Puts, Gaulin, & Verdolini, 2006) where men (ages 18–24) were recorded as they spoke to a male competitor.
In Study 3, F0 and Df were both raised and lowered independently using the procedures described above. An experimental manipulation of 1.5 JND was chosen because, from our data, 100% of subjects should perceive differences between voices raised and lowered by this amount in either F0 or Df. At the same time, this manipulation remains close to the JND, where small increments are maximally likely to have comparable effects on perception (Marks & Gescheider, 2002). For F0 manipulation, voices were raised and lowered by 1.8 semitones (1.5 JND) from baseline levels, and for Df manipulation, formant structure was shifted up and down by 6% (1.5 JND), without affecting tempo.
Thus, from each of the 30 original voices, five versions were produced: unmanipulated, raised F0, lowered F0, raised Df, and lowered Df, for a total of 150 voice recordings. These recordings were distributed into five stimulus sets of 30 recordings, each set comprising 6 raised F0, 6 lowered F0, 6 raised Df, 6 lowered Df, and 6 unmanipulated recordings and only one version of each of the 30 voices. Thus, no subject heard two versions of the same voice.
Each male subject (n=42) was seated at a computer station, where a program instructed him on the tasks to be performed, played through headphones the audio stimuli to be rated, and recorded his choices. Subjects rated 30 voices (one stimulus set) on whether each speaker was likely to be able to win physical fights (physical dominance) and whether he was likely to be a respected leader (social dominance). Physical dominance was assessed by selecting from a 10-point scale with endpoints labeled strongly agree and strongly disagree below the statement: “If this man got in a fistfight with an average male undergraduate student, this man would probably win” (Puts, Gaulin, & Verdolini, 2006). Subjects assessed social dominance by selecting from a 10-point scale with endpoints labeled extremely dominant and extremely submissive underneath Mueller and Mazur's (1997) description: “a dominant person tells other people what to do, is respected, influential, and often a leader; whereas submissive people are not influential or assertive and are usually directed by others.”
To eliminate the influence of any order effects on relationships between voice manipulations and dominance ratings, we presented the 30 stimuli in a different random order for each subject, and whether a given subject always rated social or physical dominance first was determined randomly.
2.3.2. Data treatment
Eight or nine subjects rated each stimulus set, and their ratings were averaged to produce mean physical and social dominance ratings for each recording. Thus, each original voice received 10 mean dominance ratings (5 acoustic variations×2 dominance types). Interrater reliability was high (Cronbach's α=.81±.08). Nevertheless, to control for any global differences in subjects' rating tendencies, we first standardized each subject's ratings (converted to z scores) before they were averaged with other subjects' ratings. Standardization did not affect the results.
Effects of F0 and Df on dominance ratings were analyzed using repeated measures ANOVA. All p values are two tailed; α = .05.
Three factors, each with two levels, were analyzed: manipulation (raised vs. lowered), acoustic measure (F0 vs. Df), and dominance type (physical vs. social). In separate Manipulation×Dominance Type repeated measures ANOVAs, both F0 and Df negatively affected dominance ratings [main effects of manipulations: F0: F(1, 29)=6.06, p=.020; Df: F(1, 29)=37.40, p<.001]. The effect of Df manipulation (partial η2=.56) was greater than the effect of F0 manipulation [partial η2=.17; three-way repeated measures ANOVA: Manipulation×Acoustic Measure interaction: F(1, 29)=4.32, p=.047].
Df affected both physical and social dominance [F(1, 29)=45.82, partial η2=.61, p<.001 and F(1, 29)=13.12, partial η2=.31, p=.001, respectively] and affected physical dominance ratings more than it affected social dominance ratings [Manipulation×Dominance Type interaction: F(1, 29)=7.92, p=.009; Fig. 1, Table 1]. Similarly, F0 manipulation tended to affect physical dominance ratings [F(1, 29)=7.66, partial η2=.21, p=.010] more than it affected social dominance ratings [F(1, 29)=1.89, partial η2=.06, p=.180], but this interaction was not statistically significant [F(1, 29)=0.92, p=.345; Fig. 1, Table 1].
In addition to replicating the finding that vocal masculinity affects dominance perceptions among men (Puts, Gaulin, & Verdolini, 2006), we report three novel results here. First, both Df and F0 independently affected attributions of dominance among men. Second, to the degree that our ±1.5 JND manipulations of Df and F0 were perceptually similar, Df appears to have a greater effect than F0 on dominance ratings. Finally, both acoustic measures tended to more strongly affect attributions of physical dominance than attributions of social dominance, although this difference was significant only for Df.
These results are likely to have ecological validity. To the extent that men have evolved to assess dominance during verbal interactions, the stimuli used in this study are realistic (men of the raters' age speaking spontaneously to a male competitor), and the within-speaker/between-rater design did not draw attention to the acoustic manipulations. Moreover, the effects of within-subject Df and F0 manipulations were observed in the presence of naturally occurring between-subject variation in speech content and extralinguistic features, such as amplitude and aspiration rate. Thus, Df and F0 probably affect perceptions of dominance among men in real interactions. This suggests a function for a deep male voice and its anatomical substrates: men's voices evolved to signal physical dominance.
A potential shortcoming of the present study concerns the distinction between physical and social dominance. Henrich and Gil-White (2001) argue persuasively that social influence can be achieved through force or force threat (what we call here physical dominance). Alternatively, deference may be freely given to individuals who possess valued qualities and who would thus be said to enjoy prestige. Although our variable “social dominance” in some ways approximates prestige (“is respected, influential, and often a leader”), in other respects, it connotes coercion (“tells other people what to do … assertive”). Thus, depending upon how subjects understood this variable, social dominance may conflate physical dominance (or simply “dominance”) and prestige. It is possible that Df affected ratings of respect, influence, and leadership only because this social status variable suggested the threat of force. On the other hand, subjects were also asked to rate physical fighting ability; hence, they may have understood the leadership variable to differ from this. If so, Df may have affected this variable because physical competitive ability was a prestigious quality to our subjects. Future research should more clearly distinguish between dominance and prestige.
4.1. What does a masculine voice advertise?
Several lines of evidence suggest a relationship between these acoustic parameters and physical competitive ability. First, both Df and F0 may correlate with body size. Some studies have found statistically significant relationships between F0 and men's height (Graddol & Swann, 1983) and weight (Evans, Neave, & Wakelin, 2006), although most have not (e.g., Collins, 2000, Kunzel, 1989, Lass & Brown, 1978, Rendall et al., 2005, Van Dommelen & Moxness, 1995). Similarly, some studies have found relationships between Df and men's height (Evans et al., 2006, Rendall et al., 2005) and weight (Evans et al., 2006, Gonzalez, 2004), although others have not [Collins, 2000 (neither height nor weight); Gonzalez, 2004 (weight but not height); Rendall et al., 2005 (height but not weight)]. Relatively small sample sizes may account for some of these discrepancies. However, if Df is developmentally constrained to more closely reflect height than is F0 (Fitch & Hauser, 1995), this could explain our observation that Df more strongly affects dominance attributions.
F0 also appears to correlate negatively with circulating androgens (Dabbs & Mallinger, 1999), which have been positively related to physical aggressiveness (Harris, 1999) and physical prowess (Clark & Henderson, 2003). More generally, it has been suggested that masculine traits such as low Df and F0, whose development or maintenance depends on high androgen levels, may be honest signals of health and vigor (Folstad & Karter, 1992). Finally, some aspects of men's voices may reflect self-perceived dominance (Gregory, 1994, Puts et al., 2006). For example, men who rate themselves as physically dominant tend to lower their F0 from baseline when competing, whereas men who rate themselves as nondominant tend to raise it (Puts, Gaulin, & Verdolini, 2006).
4.2. Selection for low male voices versus high female voices
Df and F0 may have evolved to exaggerate the appearance of size in men (Fitch & Giedd, 1999, Morton, 1977). This could occur if there were a strong enough correlation between voice and physical prowess for selection to favor deference to masculine voices. However, current utility is insufficient evidence that a trait is an adaptation; there must also be evidence that the trait was modified for this function (West-Eberhard, 1992). Unfortunately, most vocal structures are soft tissue and would not fossilize, and scant data exist on sex differences in primate vocal anatomy. Thus, evolutionary trends in vocal anatomy cannot easily be established by paleontological or comparative means at present.
On the other hand, ontogeny offers clues. Precipitous pubertal changes suggest that men's voices have been modified to sound deeper, with developmental events added to elongate the vocal tract and lengthen and thicken the vocal folds relative to overall body size. Of course, it is also possible that such pubertal changes occurred in both sexes ancestrally and that females subsequently lost these developmental events. This could occur, for example, due to selection for neotenic features in women (e.g., Jones, 1995), which might increase their apparent residual reproductive value. Indeed, men may prefer higher female voices (Collins & Missing, 2003). However, sexual selection tends to be stronger among males in mammals generally and in humans in particular (Daly & Wilson, 1988).
4.3. Male contests versus female choice
Men's voices may have been modified over human evolution to sound deeper, but if so, was the function to increase physical dominance among men or attractiveness to women? In fact, F0 and Df seem to signal dominance more effectively than they increase attractiveness. Based on a comparison of effect sizes, simultaneous manipulation of Df and F0 affected men's judgments of physical dominance nearly 15 times more than they affected fertile–menstrual phase women's judgments of sexual attractiveness (Puts, Gaulin, & Verdolini, 2006). This does not necessarily imply that men's voices affect mating success primarily through dominance. Voice attractiveness could have a larger effect than voice dominance on men's mating success. However, with age and sociosexuality (attitudes toward uncommitted sex) statistically controlled, physical dominance ratings of a man's spontaneous speech significantly predicted his number of sex partners over the last year, but sexual attractiveness ratings did not (Puts, Gaulin, Verdolini, & Hodges, 2006).
Although more work is needed, these data suggest that men's voices may have evolved as dominance signals, and women secondarily evolved preferences for aspects of men's voices that conveyed information about mate quality. It has been suggested, for example, that men's voices signal genetic quality (Feinberg et al., 2006, Hughes et al., 2002, Puts, 2005), including perhaps, heritable physical competitive ability. Once evolved, female preferences might then have become complementary selection pressures on men's voices.
Clark & Henderson, 2003 1.Behavioral and physiological responses to anabolic–androgenic steroids. . Neuroscience and Biobehavioral Reviews. 2003;27(5):413–436.
Cleveland, 1977 2.Acoustic properties of voice timbre types and their influence on voice classification. . Journal of the Acoustical Society of America. 1977;61(6):1622–1629.
Collins, 2000 3.Men's voices and women's choices. . Animal Behaviour. 2000;60:773–780.
Collins & Missing, 2003 4.Vocal and visual attractiveness are related in women. . Animal Behaviour. 2003;6:997–1004.
Dabbs & Mallinger, 1999 5.High testosterone levels predict low voice pitch among men. . Personality and Individual Differences. 1999;27:801–804.
Daly & Wilson, 1988 6.Homicide. . New York: Aldine de Gruyter; 1988;.
Evans et al., 2006 7.Relationships between vocal characteristics and body size and shape in human males: An evolutionary explanation for a deep male voice. . Biological Psychology. 2006;72(2):160–163.
Feinberg et al., 2006 8.Menstrual cycle, trait estrogen level, and masculinity preferences in the human voice. . Hormones and Behavior. 2006;49:215–222.
Feinberg et al., 2005 9.Manipulations of fundamental and formant frequencies affect the attractiveness of human male voices. . Animal Behaviour. 2005;69:561–568.
Fitch & Giedd, 1999 10.Morphology and development of the human vocal tract: A study using magnetic resonance imaging. . Journal of the Acoustical Society of America. 1999;106(3 Pt 1):1511–1522.
Fitch & Hauser, 1995 11.Vocal production in nonhuman primates: Acoustics, physiology and functional constraints on ‘honest’ advertising. . American Journal of Primatology. 1995;37:191–219.
Folstad & Karter, 1992 12.Parasites, bright males and the immuno-competence handicap. . American Naturalist. 1992;139:603–622.
Gaulin & Boster, 1985 13.Cross cultural differences in sexual dimorphism: Is there any variance to be explained. . Ethology and Sociobiology. 1985;6:193–199.
Gonzalez, 2004 14.Formant frequencies and body size of speaker: A weak relationship in adult humans. . Journal of Voice. 2004;32:277–287.
Graddol & Swann, 1983 15.Speaking fundamental frequency: Some physical and social correlates. . Language and Speech. 1983;26(Pt 4):351–366.
Gregory, 1994 16.Sounds of power and deference: Acoustic analysis of macro social constraints on micro interaction. . Sociological Perspective. 1994;37:497–526.
Harris, 1999 17.Review and methodological considerations in research on testosterone and aggression. . Aggression and Violent Behavior. 1999;4:272–291.
Henrich & Gil-White, 2001 18.The evolution of prestige: Freely conferred deference as a mechanism for enhancing the benefits of cultural transmission. . Evolution and Human Behavior. 2001;21(3):165–196.
Hughes et al., 2002 19.The sound of symmetry: Voice as a marker of developmental instability. . Evolution and Human Behavior. 2002;23:173–180.
Jones, 1995 20.Sexual selection, physical attractiveness, and facial neoteny: Cross-cultural evidence and implications. . Current Anthropology. 1995;36:723–748.
Kunzel, 1989 21.How well does average fundamental frequency correlate with speaker height and weight?. . Phonetica. 1989;46(1–3):117–125.
Lass & Brown, 1978 22.Correlational study of speakers' heights, weights, body surface areas, and speaking fundamental frequencies. . Journal of the Acoustical Society of America. 1978;63:1218–1220.
Lee et al., 1999 23.Acoustics of children's speech: Developmental changes of temporal and spectral parameters. . Journal of the Acoustical Society of America. 1999;105(3):1455–1468.
Marks & Gescheider, 2002 24.Psychophysical scaling. . In: Pashler H, Wixted J editor. Stevens' handbook of experimental psychology (3rd ed., Vol. 4: Methodology in experimental psychology, pp. 91–138). Hoboken, NJ: John Wiley & Sons; 2002;.
Morton, 1977 25.One the occurrence and significance of motivation—Structural rules in some bird and mammal species. . American Naturalist. 1977;111:855–869.
Mueller & Mazur, 1997 26.Facial dominance in Homo sapiens as honest signaling of male quality. . Behavioral Ecology. 1997;8:569–579.
Puts, 2005 27.Mating context and menstrual phase affect women's preferences for male voice pitch. . Evolution and Human Behavior. 2005;26:388–397.
Puts et al., 2006 28.Is low voice pitch a male dominance display?. . In: Paper presented at the 18th Annual Meeting of the Human Behavior and Evolution Society, Philadelphia, PA. 2006, June 7–11;.
Puts et al., 2006 29.Dominance and the evolution of sexual dimorphism in human voice pitch. . Evolution and Human Behavior. 2006;27:283–296.
Rendall et al., 2005 30.Pitch (F0) and formant profiles of human vowels and vowel-like baboon grunts: The role of vocalizer body size and voice-acoustic allometry. . Journal of the Acoustical Society of America. 2005;117(2):944–955.
Smith et al., 2005 31.The processing and perception of size information in speech sounds. . Journal of the Acoustical Society of America. 2005;117(1):305–318.
Titze, 2000 32.Principles of voice production. . Iowa City, IA: National Center for Voice and Speech; 2000;.
Van Dommelen & Moxness, 1995 33.Acoustic parameters in speaker height and weight identification: Sex-specific behaviour. . Language and Speech. 1995;38(Pt 3):267–287.
West-Eberhard, 1992 34.Adaptation. . In: Keller EF, Lloyd EA editor. Keywords in evolutionary biology (pp. 13–18). Cambridge: Harvard University Press; 1992;.
a Neuroscience Program, Michigan State University, East Lansing, MI 48824, USA
b Department of Anthropology, Center for Evolutionary Psychology, University of California, Santa Barbara, Santa Barbara, CA 93117, USA
c Department of Psychology, Michigan State University, East Lansing, MI 48824, USA
© 2007 Elsevier Inc. All rights reserved.