Abstract
According to current research, ratings of facial and vocal masculinity/femininity are often correlated which suggests that voices and faces may convey backup signal of common underlying quality. A number of quantifiable vocal and facial features has been shown to significantly corresponds with ratings of masculinity/femininity and clearly differ intersexually.
Here we focus on intrasexual variation in facial shape dimorphism assessed by geometric morphometrics (GM), mean fundamental frequency, and formants in voice. To our knowledge no study tested the level of correspondence between measured (in contrast to perceived) sexual dimorphism in human faces and voices.
Also little is known about the relationship between facial and vocal traits in different populations. Here we search for possible association between facial and vocal sexually dimorphic traits in four samples, namely from Brazil, Cameroon, Czech, Namibia.
GM scores of sexual dimorphism were computed from standardised facial photographs. F0 and formants were computed from short (population) standardized utterances, using PRAAT autocorrelation method for F0 and Burg method for first four formants.
Settings recommended by PRAAT authors were utilized. Formants were aggregated into single variable - formant position (Pf; Puts et al., 2012).
Due to non-normal distribution of several variables we applied Spearman rank correlations, and tested for correlation between GM shape scores of facial masculinity/femininity and mean fundamental frequency (F0) and mean formant position (Pf) of vocal recordings. GM, F0, and Pf was assessed separately for each sample.
F0 and Pf mainly because of differences in uttered texts. After Bonferroni correction none of 16 performed tests was significant.
Closest to significance was association of F0 and GM in Brazilian women (rs = -.229, p = .109). Our results suggest that facial and vocal masculinity/femininity are more complex phenomena, and/or they signal different qualities.