1. Introduction

There is growing evidence that the sound of a person's voice may convey important information. Studies have shown that listeners who hear voice samples can infer the speaker's socioeconomic status (Ellis, 1967, Harms, 1963), personality traits (Allport & Cantril, 1934, Zuckerman & Driver, 1989), and emotional and mental state attributes related to deception (Ekman et al., 1976, Streeter et al., 1977). Listeners exposed to voice samples are also capable of estimating the age, height, and weight of speakers with the same degree of accuracy achieved by examining photographs (Krauss et al., 2002, Lass & Colt, 1980, Lass & Davis, 1976). Independent raters are also capable of matching a speaker's voice with the person's photograph over 75% of the time (Krauss et al., 2002). Hughes, Harrison, and Gallup (2002) found that individuals with symmetrical morphological traits were rated as having more attractive voices, and as deviations from bilateral symmetry increased, ratings of voice attractiveness decreased.

Vocal development is influenced and modified by activational sex hormones during adolescence. Estrogen and progesterone shape the mature female voice, while testosterone modifies the male voice (Abitbol, Abitbol, & Abitbol, 1999) The same hormones that affect voice have also been implicated in the development of sex-specific body configuration features. For instance, waist-to-hip ratios (WHR) in females become more accentuated during puberty, and WHR is an indicator of a female's hormonal profile (increased estrogen) and reproductive status (Singh, 1993). Similarly, during puberty, the male's shoulders broaden, increasing his shoulder-to-hip ratio (SHR). Broad shoulders and narrow hips appear to be shaped by testosterone (Kasperk et al., 1997). Because both voice and WHR/SHR are affected by sex hormones, we examined whether ratings of voice attractiveness were related to variation in SHR and WHR.

WHR and SHR not only influence judgments of attractiveness (Dijkstra & Buunk, 2001, Singh, 1993), but also, variation in these traits predicts different aspects of sexual behavior in both males and females (Hughes & Gallup, 2003). Women with low WHR (smaller waist compared with the hips) are rated as being more attractive (Beck et al., 1976, Singh, 1993), report having sex at an earlier age, and report having more sex partners (Hughes & Gallup, 2003, Mikach & Bailey, 1999). Men with high SHR (larger shoulders and smaller hips) are likewise judged to be more attractive (Dijkstra & Buunk, 2001) and are also more sexually experienced (Hughes & Gallup, 2003). Therefore, we also examined the relationship between ratings of voice attractiveness and different features of sexual behavior.

2. Methods

3. Results

3.1. Voice and morphology

Male SHRs (M=1.15±0.076, range 1.01–1.47) exceeded female SHRs (M=1.04±0.075, range 0.84–1.19; t=8.69, p<.01). Male WHRs (M=0.86, S.D.=0.75, range 0.74 to 1.18) also exceeded female WHRs (M=0.78, S.D.=0.61, range 0.64 to 0.97; t=6.81, p<.01). Likewise, male BMI (M=25.35±4.06) was higher than female BMI (M=23.87±4.24; t=2.15, p<.05). Correlations between opposite- and same-sex ratings of voice attractiveness and morphological features are presented in Table 1, Table 2. Opposite-sex ratings of voice attractiveness were positively correlated with SHR (r=.503 p<.01) in males and negatively correlated with WHR (r=−.376, p<.01) in females. There was no relationship between voice attractiveness and BMI in either males or females. Controlling for the other morphological features had little or no effect on these outcomes. The partial correlation between voice attractiveness and SHR in males was r=.467 (p<.01), whereas for females, the partial correlation between voice attractiveness and WHR was r=−.375 (p<.01). Partial correlations between voice attractiveness ratings and BMI when controlling for SHR, WHR, as well as age were nonsignificant in both males and females.

Table 1. Pearson correlations (sample size) between voice attractiveness ratings, WHR, SHR, BMI, and various measures of sexual behavior in males

	Males	Opposite-sex ratings of voice attractiveness	Same-sex ratings of voice attractiveness	WHR	SHR	BMI	Age of first masturbation	Age of first sex	Number of sex partners	Number of times being an EPC partner
	WHR	.150 (70)	.174 (70)
	SHR	.503**(70)	.375** (70)	.302* (70)
	BMI	.024 (70)	−.093 (70)	.575** (70)	.118 (70)
	Age of first masturbation	−.101 (36)	−.219 (36)	.071 (36)	−.233 (36)	.105 (36)
	Age of first sex	−.410* (34)	−.273 (34)	.159 (34)	−.419* (34)	−.061 (34)	.312 (30)
	Number of sex partners	.359* (43)	.327* (43)	.187 (43)	.533** (43)	.161 (43)	−.357* (36)	−442** (34)
	Number of EPC partners	.400** (43)	.356* (43)	.150 (43)	.491** (43)	.132 (43)	−.289 (36)	−.373*(34)	.781** (43)
	Number of times being an EPC partner	.317* (43)	.261 (43)	.181 (43)	.449** (43)	.036 (43)	−.143 (36)	−.381* (34)	.548** (43)	.551** (43)

* p<.05 (two tailed). ** p<.01 (two tailed).

Table 2. Pearson correlations (sample size) between voice attractiveness ratings, WHR, SHR, BMI, and various measures of sexual behavior in females

	Females	Opposite-sex ratings of voice attractiveness	Same-sex ratings of voice attractiveness	WHR	SHR	BMI	Age of first masturbation	Age of first sex	Number of sex partners	Number of times being an EPC partner
	WHR	−.376**(76)	−.021 (76)
	SHR	−.107 (76)	.192 (76)	.189 (76)
	BMI	.000 (76)	−.011 (76)	.363** (76)	−.514** (76)
	Age of first masturbation	.043 (17)	−121 (17)	.280 (17)	−.182 (17)	.265 (17)
	Age of first sex	−.397 (28)*	−.171 (28)	.376* (28)	−.345 (28)	.575** (28)	.264 (14)
	Number of sex partners	.491** (34)	.409* (34)	−.349* (34)	−.046 (34)	−.107 (34)	−.216 (17)	−.376* (28)
	Number of EPC partners	.374* (34)	.295 (34)	−347* (34)	−.019 (34)	−.122 (34)	.022 (17)	−.242 (28)	.490** (34)
	Number of times being an EPC partner	.353* (34)	.401* (34)	−.339* (34)	−.023 (34)	−.134 (34)	−.025 (17)	−.117 (28)	.566** (34)	.612** (34)

* p<.05 (two tailed). ** p<.01 (two tailed).

3.2. Morphology and sexual behavior

Descriptive statistics for self-reported measures of sexual behavior are presented in Table 3. As shown in Table 1, SHR for males was negatively correlated with reported age of first sexual intercourse and positively correlated with reported number of sex partners, number of EPC partners, and number of people they had intercourse with who were involved in another relationship. For females, WHR was positively correlated with reported age of first sexual intercourse, and negatively correlated with the number of reported sex partners, EPC partners, and partners they had intercourse with that were involved in another relationship (see Table 2). Whereas BMI was not correlated with any of the measures of sexual behavior in males, BMI was correlated with age of first sexual intercourse for females (r=.575, p<.01), suggesting that increases in body mass postpone the onset of heterosexual intercourse in females.

Table 3. The mean and standard deviation for the sexual behavior measures in males and females

		Males			Females
		n	Mean	S.D.	n	Mean	S.D.
	Age of first masturbation	36	13.16	1.78	17	13.82	2.98
	Age of first sex	34	16.94	1.37	28	16.82	1.68
	Number of sex partners	43	4.44	6.61	34	2.94	2.95
	Number of EPC partners	43	0.53	1.22	34	0.23	0.55
	Number of times being an EPC partner	43	0.19	0.045	34	0.21	0.48

4. Discussion

Our findings show that ratings of voice attractiveness can be used to predict significant variation in sexually dimorphic body configuration. Females with smaller WHRs and males with larger SHRs had voices that were consistently rated as more attractive. For females, WHR conveys information about hormonal profile, reproductive maturity, fecundity, and health (for a review, see Singh, 1993). For males, SHR correlates positively with beta-lipoproteins, hormones that are related to testosterone and muscle development (Evans, 1972), and higher androgen levels in males may contribute to the development of sex-related differences in skeletal morphology and muscle mass at puberty (Kasperk et al., 1997). Because the sex hormones that influence the emergence of sexually dimorphic configurations of SHR and WHR are the same as those that drive the emergence of sex differences in vocal structure during adolescence (Abitbol et al., 1999), there may be a fairly direct hormonal link between these features. Whereas male SHR was correlated with both opposite- and same-sex ratings of voice attractiveness, female WHR was only predicted by opposite-sex voice attractiveness ratings.

Some investigators contend that WHR is confounded with body mass, especially among females (e.g., Wetsman & Marlowe, 1999). While our data show that WHR is related to voice attractiveness in females, there was no evidence for an association between BMI and voice attractiveness for either sex. Some findings show that height and weight can be estimated from voice samples (Krauss et al., 2002, Lass & Colt, 1980, Lass & Davis, 1976). But when it comes to ratings of voice attractiveness, our results show that height and weight, as scaled by BMI, are not related. Although Collins and Missing (2003) found that vocal attractiveness was negatively related to female “body size,” our data suggest that the relationship between voice attractiveness and body features is driven by the variation in WHR, not BMI. Unlike WHR and SHR, the only significant correlation between BMI and any of the measures taken in our study was age of first sexual intercourse among females. As BMI for females increased, there was a corresponding increase in the reported age of first sexual intercourse. Therefore, it would appear that increases in BMI retard or delay the onset of heterosexual intercourse among females.

We also found associations between voice attractiveness ratings and different features of self-reported sexual behavior. Both males and females whose voices were rated as attractive had sex at an earlier age, had more sexual partners, more EPC partners, and more sexual partners that were involved in a relationship with another person. It is interesting that voice attractiveness ratings by members of the opposite sex were better predictors of sexual behavior than ratings by members of the same sex. Aside from Wilson (1984), who noted that lower voiced male opera singers were more inclined to have sexual affairs with fellow singers, our findings are the first to empirically implicate the existence of a relationship between voice and sexual behavior.

Individuals with attractive voices are perceived more favorably and as having more desirable personality characteristics (Zuckerman & Driver, 1989). Furthermore, the higher the ratings of voice attractiveness, the more the speaker is judged to be similar to the rater and the more the rater would like to affiliate with the speaker (Miyake & Zuckerman, 1993). This “vocal attractiveness stereotype” (Zuckerman & Driver, 1989, Zuckerman et al., 1990) may promote sexual opportunities. Although Zuckerman and Driver (1989) did not find an effect, Collins and Missing (2003) report a substantial correlation between ratings of voice attractiveness and of facial attractiveness in women. Therefore, since ratings of facial attractiveness predict semen quality in males (Soler et al., 2003) and longevity in both males and females (Henderson & Anglin, 2003), voice attractiveness may be an indicator (albeit indirect) of other fitness-related features as well.

Our findings replicate those of Hughes and Gallup (2003) concerning the relationship between various measures of sexual behavior and body configuration. Reported age of first sex, number of sex partners, number of EPC partners, and number of sexual partners who were involved in another relationship were all significantly correlated with SHR in males and with WHR in females (see Table 1, Table 2). However, unlike fluctuating asymmetry (Hughes et al., 2002), the relationship between voice attractiveness and body configuration in females was influenced by the sex of the rater.

It is interesting that voice attractiveness was a better predictor of sexual behavior in females than WHR was. The best predictor of promiscuity (as measured by number of sexual partners) in males was SHR, whereas in females, it was voice attractiveness. As shown in Table 2, the correlations between opposite-sex ratings of voice attractiveness and sexual behavior in females were consistently higher than those found for WHR and sexual behavior. However, just the opposite was true for males (see Table 1). Among males, SHR was a better predictor of sexual behavior than voice was. For example, whereas opposite-sex voice attractiveness accounted for 13% of the variance in promiscuity among males, SHR accounted for 28% of the variance in male promiscuity. The same sex differences held for the relationship between voice and body configuration. Male SHR accounted for more variance (25%) in opposite sex ratings of voice attractiveness than did female WHR (14%).

Our findings suggest that during human evolutionary history, voice may have been an important parameter of mate choice, particularly at night, when vision was compromised (for details, see Hughes et al., 2002). Not only does voice map onto language as our principal mode of communication, but it appears that irrespective of content. The mere sound of an individual's voice also provides important information about a host of biologically relevant features such as fluctuating asymmetry, body configuration, and sexual behavior.