Vocal modulation in human mating and competition

Abstract

The human voice is dynamic, and people modulate their voices across different social interactions. This article presents a review of the literature examining natural vocal modulation in social contexts relevant to human mating and intrasexual competition. Altering acoustic parameters during speech, particularly pitch, in response to mating and competitive contexts can influence social perception and indicate certain qualities of the speaker. For instance, a lowered voice pitch is often used to exert dominance, display status and compete with rivals. Changes in voice can also serve as a salient medium for signalling a person's attraction to another, and there is evidence to support the notion that attraction and/or romantic interest can be distinguished through vocal tones alone. Individuals can purposely change their vocal behaviour in attempt to sound more attractive and to facilitate courtship success. Several findings also point to the effectiveness of vocal change as a mechanism for communicating relationship status. As future studies continue to explore vocal modulation in the arena of human mating, we will gain a better understanding of how and why vocal modulation varies across social contexts and its impact on receiver psychology.

This article is part of the theme issue ‘Voice modulation: from origin and mechanism to social impact (Part I)’.

1. Introduction

Humans are intensely communicative primates, and vocalization is a fundamental component of human communication. Although lexical content and syntactic structure define language, the prosodic features of speech, such as intonation and pitch modulation, may be equally salient (e.g. [1]). The dynamics of human mating are, therefore, likely to be clarified by an understanding of how speakers modulate their voices across mating contexts, and how these modulations influence the perceptions and responses of listeners. The ability to modulate the voice can function to communicate romantic intentions, attraction, mate value, formidability and social status, for example.

Numerous empirical studies have investigated how artificial manipulation of vocal acoustic parameters affects perceptions, especially how the alteration of fundamental frequency using specialized software can shape assessments of speaker attractiveness and formidability (for reviews see [2,3]). Fundamental frequency (f_o) is the rate of vocal fold vibration during phonation and the acoustic parameter closest to what is perceived as pitch. Here, we focus on a smaller but rapidly growing number of studies that have examined speakers' natural modification of their voices, often without conscious control or awareness [4], within the contexts of mating and intrasexual competition. Examining natural modification of the voice as opposed to artificial manipulations has high ecological validity essential for understanding the communicative functions of voice during mate selection and intrasexual competition. A summary is needed of this growing evidence regarding how people naturally regulate their voices to advertise or exaggerate traits and states relevant to sexual selection.

(a) . Theoretical background: why does the voice respond to social context?

At a proximate level, voice modulation across social contexts involves both the motor cortex [5] and regions of the brain involved in social processing, including the medial prefrontal cortex, superior temporal sulcus and precuneus [6]. From a developmental perspective, the capacity to formulate vocal expression and inflection is evident early in life and is seen when infants babble [7]. These critical elements are not necessarily learned [8,9], although older children can effectively be taught to improve their ability to use vocal expression and volume to enhance communication skills [10]. Some of the human capacity to modulate acoustic features of vocalizations, such as f_o, is also shared with a number of non-human primates (e.g. [11]). However, to provide a framework for understanding vocal modulation, it is helpful to consider why, at a functional level, particular acoustic parameters may be linked to certain social contexts and associated emotional and motivational states [12,13].

In some cases, acoustic modulations may represent non-functional byproducts of physiological reactions to social contexts, such as ‘fight or flight’ endocrine stress responses [14] and peripheral muscle tension [15]. For example, speaking and breathing compete for some of the same physiological resources. Changes in respiratory rate and tidal volume that accompany sympathetic arousal or physical exertion may influence multiple acoustic features, including breathiness, mean f_o and variation in f_o (sometimes measured by the standard deviation in f_o across the utterance, f_o − s.d.) [14,16,17]. In these cases, acoustic parameters may constitute cues of motivational state, in that they convey information but have not evolved for that function.

In other cases, acoustic correlates of emotional and motivational states may have evolved for the function of transmitting information. For example, the high amplitude vocalizations that often accompany rage may function to signal propensity towards physical aggression because high amplitude vocalizations are difficult to ignore and require abundant energy for their production. By contrast, reducing amplitude and increasing breathiness may serve to convey a lack of aggressive intent. In such cases, acoustic changes would constitute signals [13,18–20].

The specific acoustic signals that evolve are likely to depend on preexisting population-level associations between acoustic parameters and vocalizer characteristics, as well as preexisting sensitivities to these acoustic parameters [21]. If a particular acoustic feature tends to predict some salient vocalizer quality, then vocalizers may leverage this association by modulating their voices to convey that quality. Knowing how acoustic parameters are related to important vocalizer characteristics at a broad scale can then help us understand how and why individuals modulate their voices across social contexts. Likewise, vocalizers may take advantage of preexisting sensitivities to manipulate the behavioural responses of receivers [22].

Thus, it is useful to consider two related questions: (i) What do various acoustic parameters indicate about a speaker? and (ii) How do these parameters influence social perceptions? For example, larger animals tend to vocalize at a lower f_o across species [23,24] and to varying but generally lesser degrees within species [25,26]. Vertebrate vocalizations can be described by the source-filter theory, in which voice production involves the independent contributions of the vibrating anatomical structures that provide the sound source and the structures that filter the sound [27,28]. In humans, a large sound source (vocal folds) and longer filter (supralaryngeal vocal tract) are associated with lower f_o and formant frequencies (resonant frequencies of the vocal tract that influence perceptions of timbre), respectively [28]. Perhaps due to the association between low vocal frequencies and large body size, mammals including humans tend to lower their vocal frequencies in contexts where it would benefit them to appear large and threatening, and to raise vocal frequencies when signalling subordination or affiliation [29–31].

Males likely evolved a relatively low f_o compared to females in some primate species due to a selective advantage of sounding larger and more formidable during the intermale aggressive competition for mates [24]. Low male f_o may be particularly likely to evolve in species in which males compete intensely for mates but can often avoid fights through threat displays [24]. Because female fitness tends to be less strongly tied to mating success in primates, females tend to experience weaker selection for exaggerated ornamentation and weaponry, such as large body size, long canine teeth and deep resonating vocalizations. Among humans, adult males speak at approximately half the f_o of females [28,32]. This association between vocal frequencies and biological sex affords the opportunity to manipulate another set of social perceptions: those related to masculinity and femininity. Lowering f_o increases perceptions of masculinity, and raising it increases perceptions of femininity [33,34]. Similarly, formant frequencies are negatively correlated with both body size and perceptions of size and dominance, are lower in men than women [33,35], and hence may also be modulated in contexts where there is a benefit to manipulating apparent size, dominance, or masculinity. Harmonics-to-noise ratio (HNR) may also be higher in women's than men's voices and influences perceptions of breathiness [36,37], which may, therefore, influence perceptions of femininity, youth and attractiveness [38–40] and influence mating opportunities [41].

The key point is that associations between acoustic parameters and characteristics of vocalizers on a broad (e.g. population) scale enable vocalizers to manipulate listeners' perceptions. Socially relevant interpersonal dimensions such as size, formidability, masculinity and femininity, as well as behavioural and relational predispositions—for instance, aggression or affiliation—can be communicated through vocal modulation. As an example, it has been shown that vocalizations with increased intensity (i.e. amplitude) and duration predicted greater perceptions of men's formidability; however, f_o measured from these intimidating ‘human roars’ predicted only perceived fighting ability but not actual fighting success in this sample [42].

This apparent mismatch between perception and actuality raises another key point: because the interests of the signaller (vocalizer) and receiver (listener) are frequently divergent, selection should favour receivers who are not easily manipulated by dishonest signals of size, formidability, mate quality and the like. Some deceptive signalling is likely to evolve, but game-theory models indicate that deception must be infrequent or costly for the signalling system to be evolutionarily stable [43–45]. Deceptive vocal modulation may be costly if the signaller pays social or physical costs when the deception is detected, for example, aggressive retribution from same-sex competitors [46]. Some vocal modulation may also be honest insofar as it is constrained by adaptive physiological responses to social context, such as increased sympathetic arousal and breathing rate that accompany certain emotional states. Such responses may affect multiple acoustic parameters. As an example, hot rage is associated with increases in mean f_o, f_o − s.d., and vocal intensity [17].

Thus, in general, we can expect vocal modulation to leverage broader associations between acoustic features on the one hand, and characteristics or motivational states of the vocalizer on the other [47], to manipulate listeners' perceptions. Such manipulation is likely to be successful in proportion to the honesty of the signal, the costs of detecting dishonesty and the extent to which the fitness interests of signaller and receiver converge. In the sections that follow, we will return to these themes as we consider empirical research on vocal modulation across social contexts related to human mating.

(b) . Vocal modulation and human sexual selection

Although a thorough consideration of human sexual selection is beyond the scope of this review, it is worth highlighting some key features, as these can help clarify how the voice is modulated across contexts related to mating. Sexual selection favours traits that win mating opportunities (e.g. the elk stag's antlers or the peacock's colourful train) and tends to be stronger in the sex that invests less in offspring and is capable of reproducing at a higher maximum rate [48,49]. In humans, as in mammals generally, males have a lower minimum obligatory investment in offspring than females and are capable of reproducing at higher rates by reproducing with multiple mates; hence, we expect sexual selection to have been more intense among our male ancestors [50–53] and thus to have played a relatively larger role in shaping male vocalizations [24,54]. Nevertheless, human males provide an extraordinary degree of parental investment compared to males of most other mammalian species, and so we can also predict an unusual degree of mating competition among our female ancestors [55]. Thus, competition has likely also shaped female vocal behaviours to some extent [56], as we discuss below.

Mating competition can take different forms, including contests, the use of force or threat of force to exclude same-sex competitors from mating opportunities, and mate choice, which favours displays, ornaments and other phenotypes that attract mates. Available evidence indicates that mate choice and contests were important in ancestral mating competition in both sexes, but that contests were relatively more important in males, and mate choice was more important in females [53,57–61]. This difference may be expected to influence the relative salience of varying contextual factors, such as the presence of potential mates or same-sex competitors, on how males and females modulate their voices.

2. Dominance and status competition

We first consider vocal modulation across contexts related to contest competition—those associated with intimidating and/or winning status among same-sex competitors. When asked to use their voices to convey several traits pertinent to mating competition, both sexes were able to modulate the perceived dominance and intelligence of their voices [62]. However, men could effectively modulate their voices to portray confidence but not sexiness, whereas women could portray vocal sexiness but not confidence. These sex-specific abilities may reflect sex differences in the importance of mate choice versus contests to ancestral mating success [55,60], as well as differences in qualities that are most desired by the opposite sex. Confidence may suggest earning potential, social power and personality characteristics related to success that women find desirable, whereas men tend to place more emphasis on physical attractiveness when finding mates [63,64].

Whether or not vocal modulation is deliberate, changes in perceived dominance and related variables such as size, formidability, masculinity and status tend to predict changes in f_o and formants. For example, when asked to sound masculine or feminine, both sexes decreased or increased their f_o and formants, respectively [65]. Canadian, Cuban and Polish speakers tended to decrease their formant frequencies and f_o when imitating large size and increase these parameters when imitating small size, with men generally modulating their voices, particularly formants, more than women did [66]. Similarly, lowering f_o was found to increase perceptions of dominance [67], authoritativeness [68] and likelihood of attacking an adversary, but not fighting ability, in a conflict situation [69]. Some evidence also indicates that these perceptual effects of vocal modulation influence real-world outcomes. When participants worked together in a problem-solving task, lowering or raising f_o predicted higher and lower eventual rank, respectively [70]. In the context of tennis competitions, players exhibited grunts that listeners used to accurately identify both their gender and the outcome of the contests (i.e. players who won contests had lower f_o grunts than the players who lost; [71]).

If vocal shifts such as lowering f_o or formants increase status outcomes, and if men and women can produce these shifts volitionally, then why do people not habitually speak in the most status-evoking voices that they can physically produce? The answer is almost certainly that this would not be beneficial. Even in the context of speaking to competitors for mates, there are many circumstances that call for signalling affiliation or deference, especially if dominance signalling risks incurring costly challenges. For example, under a sexual priming condition, low-f_o male voices elicited aggressive cognitions and intent in men who perceived themselves to be more dominant and stronger [72]. Males who lower f_o when speaking to a competitor could thus incur dangerous retribution. As in other mammalian species, human vocalizations produced in an aggressive context seem to enhance the expression of threat and signal the physical formidability of the vocalizer, whereas vocalizations produced in a distressed or submissive context are often displayed so as to minimize perceived threat [73].

Indeed, several studies have shown that the voice is modulated in contexts related to status and dominance [74]. Speakers seem to adjust their vocal parameters according to their own social status, often in relation to the perceived social status of a listener. For example, male university students who believed that they were a better fighters than their competitors in a dating game scenario lowered their f_o when addressing their rival, whereas men who believed they were a worse fighter raised it [54]. Similarly, both men and women who judged themselves to be more dominant or prestigious during a mock job interview lowered both mean f_o and f_o variability, and those who believed they were subordinate in these forms of status increased their f_o in response to more dominant and prestigious targets [75]. In addition, professional academics lowered their vocal frequencies when providing expert advice compared to providing information that was common knowledge [68], again indicating that relative authority influences voice modulation.

In many of these studies, participants rated their interlocutor's and their own social status after the interaction, raising the possibility that participants gauged their relative status retrospectively by monitoring their own voice shifts. However, this does not explain why speakers modulated their voices in the first place. Status posturing through voice may occur without conscious awareness and may be exhibited because it is reinforced by the receiver's response to it. Voice modulation thus represents a socially salient means by which relative status can be signalled and negotiated across social contexts and even during a single interaction. Collectively, these findings support the idea that transient vocal changes play a vital role in assessing and establishing status relationships during intrasexual competition. It is important to highlight the relative paucity of evidence regarding women's vocal modulation within intrasexual competitive contexts, however. Intrasexual mating competition may have been less intense among our female ancestors compared to males [41,60] but was nevertheless present [56,57,76]. Further investigations should clarify how women modulate their voices during intrasexual competitive situations. For men, evidence suggests that contest competition was an important selection pressure on vocal behaviours [77].

3. Mate attraction

Researchers have also begun to explore patterns of voice modulation in contexts related to mate attraction. A few examinations have focused on the conscious and deliberate production of attractive vocalizations. For example, participants asked to produce a ‘sexy voice’ slowed their speech and lowered their f_o [62,78]. Women decreased f_o more than men did, displayed greater vocal hoarseness, and, as noted above, had greater success at producing sexier voices than their baseline voice [62]. However, other research found that voice samples obtained after participants were instructed to deliberately alter their pitch in a sex-typical manner (i.e. lowering in men, raising in women) were not perceived as more attractive than baseline recordings, whereas when participants intentionally altered their pitch in a sex-atypical manner (i.e. raising in men, lowering in women), perceived attractiveness tended to decrease [67]. An f_o that deviates too far from the average within each sex's range could indicate pathologies [79], and deliberately modulating f_o could also alter other acoustic parameters in a way that decreases attractiveness, perhaps in combination with f_o changes. Thus, instruction to deliberately raise or lower the pitch of one's voice appears to yield different results with respect to perceived attractiveness than instruction to make one's voice sound sexier. Seemingly, alteration of acoustic features other than pitch is important for producing an attractive-sounding voice, and it has been shown that a unique constellation of acoustic features contributes to perceived vocal attractiveness (see [38]).

Beyond responses to explicit instruction, the heightened physiological arousal that accompanies interaction with an attractive target [80] is likely to influence both verbal and nonverbal vocal expression. When recalling a first interaction with an attractive person or potential romantic partner, both sexes reported faster speech and reduced ability for clear verbal expression, and women particularly reported higher pitch and unsteady tone of voice [81].

(a) . Female voice modulation when speaking to potential mates

Results from studies that examined unsolicited natural vocal modulation during speed-dating events largely matched these recollections for women and provide further insight regarding how individuals modulate their voice during mate selection. Women used a higher f_o and f_o variability (f_o − s.d.) towards men they preferred as potential mates in speed dates [82]. However, women lowered their f_o when interacting with men who were chosen by both them and most other women in the dating event [82]. In experiments in which stimuli were selected based on prior attractiveness ratings, women also lowered their f_o when speaking to a more attractive male target depicted in a facial image [80], but not in a soundless video [83]. Likewise, women tended to lower f_o − s.d. when speaking to a generally desired speed-dating partner [82], but not to an attractive male in a soundless video [83]. The results of these studies are mixed but suggest that women often decrease f_o and f_o − s.d. when speaking to a potential mate of high mate value as assessed by others, even though they may increase these parameters when speaking to a man whom they personally select. Because greater f_o and f_o − s.d. have been associated with perceptions of female speakers' affection towards their conversational partner [84] and with empathy [85], one can speculate that women sometimes raise these parameters when they find a potential mate interesting and attainable but lower these parameters to avoid ‘showing their hand’ to attractive males by signalling excessive interest. Data are equivocal regarding whether higher f_o female voices are generally preferred [24,56,86,87], but lowering f_o could also signal seduction and proceptivity [80].

Women's voice modulation in contexts of mate attraction appears to be effective. Recordings of speech directed towards attractive versus unattractive targets were rated as more attractive by independent listeners [80], even when listeners spoke a different language and when several acoustic properties were removed from the voices by low-pass filtering, rendering the speech unintelligible [83]. The increased attractiveness of speech directed towards attractive targets seemed to be related in part to f_o.

(b) . Male voice modulation when speaking to potential mates

In these studies examining speed-dating [82] and responses to videos [83], men tended to lower their minimum, but not mean, f_o when speaking to attractive women. Although there was no effect of target attractiveness on f_o − s.d. in these studies, a more general effect has been observed in which men lowered their f_o − s.d. from a baseline recording when they spoke to a potential date [88]. This change was more pronounced among men who rated themselves as relatively better fighters compared to their competitor in the experiment. Whereas f_o − s.d. is a poor predictor of men's attractiveness judgements made by women [24,88], men who spoke with a lower f_o − s.d. when describing themselves to a potential date reported a greater number of past-year sexual partners after controlling for age and sociosexuality [77]. Another study found that a lower f_o − s.d. attenuated the negative effects of vocal roughness and breathiness on men's number of recalled sex partners [41]. A stable f_o could reflect the confidence that men with greater formidability and higher mating success have when placed in a competitive mating context.

Voice modulation in contexts of mate attraction also appears to be effective for males. Men who exercised greater vocal modulation and gradually deepened their voices during conversations with unfamiliar women in a simulated dating scenario were more successful at getting future dates [89]. Recordings of men's speech directed towards physically attractive versus unattractive female targets were rated as sounding more attractive by independent listeners [80,83], independent of language and removal of several acoustic properties by low-pass filtering [83]. As with women's voices, the increased attractiveness of men's speech directed towards attractive targets seems to be related in part to f_o [80,83]. During spontaneous mixed-sex dating conversations, women did not generally judge male speakers with a lower f_o as more attractive but rather found male speakers who talked lower within their own natural f_o range to be attractive [90]. The authors concluded that it may not be the actual or relative f_o of the voice itself that speakers perceived as attractive but rather the fact that a speaker's voice indicated attraction towards their conversational partner. Alternatively, perhaps lowering f_o is an indicator of the ambivalence that tends to characterize someone who perceives themselves to be relatively high in mate value. Indeed, a lower male f_o is sometimes [24,33,61] but not always [91,92] found to increase attractiveness to women.

(c) . Vocal synchrony

In general, unacquainted individuals actively match their f_o and temporal speech patterns during conversations [1,93,94]. In a mock speed-dating event, women did not lower their f_o when speaking to a male ‘suitor’ whose f_o had been experimentally lowered versus raised [95]. However, another speed-dating experiment involving mixed-sex dialogues of heterosexual singles showed that speakers converged over time in both voice register and vocal range [96]. The degree of pitch convergence was dependent upon perceived attractiveness and likability of their conversational partner. In fact, speakers in a dyadic interaction could show prosodic entrainment at different degrees, and it was possible that entrainment could be one-sided depending upon each speaker's perception of their interlocutor.

Displays of vocal synchrony between human mates seem to parallel patterns seen in avian mate choice [93], and non-vocal behavioural synchrony during human courtship [97]. Vocal matching in songbirds facilitates mate attraction [98]. Indeed, higher levels of synchronicity in the speech patterns of close romantic partners were found compared to stranger dyads [94]. However, it is important to note that stranger dyads showed a steady increase in vocal convergence as their conversation exchange progressed. In comparison, close partner vocal cues (as indexed by mean f_o) did not converge; rather they were continuously correlated throughout the conversation [99]. Even in same-sex pairings, both men and women were able to learn to synchronize their temporal speech rate to their conversational partner with equal skill [93]. Vocal synchrony may signal commonality, attentiveness and affiliation to a conversational partner. The ability to display vocal synchrony with same-sex conspecifics could also signal these qualities, as well as capacity for cooperation, to potential mates [93].

4. Relationship status

Vocal modulation may be important not only in relationship formation but also in relationship maintenance. Romantic partners tend to use prosodic exaggeration, or ‘loverese’, when speaking to one another [100]. This vocal pattern is similar to infant-directed speech, formerly referred to as ‘motherese’ [101], as well as pet-directed speech [102], which involves exaggerated vocal intonation, higher pitch, broader pitch range, slower rate and longer pauses [102,103]. This exaggerated vocal profile appears to communicate a level of tenderness or care towards the target.

Voice samples obtained from conversations between couples were perceived differently by independent judges than samples obtained from other conversers. Farley, Hughes and LaFayette [104] recorded phone conversations between heterosexual participants and their newly in-love romantic partner and separate conversations with a same-sex friend. Vocal clips of these recordings were presented to independent raters, but the content of the conversation was obscured using software to create paralanguage vocalizations. This technique retained certain voice qualities such as prosody, f_o, intensity and intonation but made the words of the speech sound muddled. The voice samples directed towards romantic partners were judged by independent raters as sounding more pleasant, sexier and reflecting greater romantic interest than vocalizations directed towards same-sex friends. Further, ratings of romantic interest based on voice were positively correlated with the degree of speakers' reported love for their romantic partner [104]. In a similar study, female speakers were perceived by raters as sounding more approachable, sincere, submissive and scatterbrained during phone conversations with their intimate male partner than with their male friend [105].

Listeners have also been found to recognize close partner vocalizations as opposed to those obtained from stranger conversations with better-than-chance accuracy from 1-min segments of conversations without verbal content [94]. Whereas the stranger partner conversations tended to have more uniform or homogeneous vocal features, close partner vocalizations carried their own set of unique characteristics. Independent listeners thought that close partner vocalizations sounded relaxed, included many vocal changes, and they perceived that the woman's tone decreased in a playful way. By contrast, stranger partner conversations entailed very little rise or fall in tones, had longer pauses, and the woman often sounded high-pitched and hurried. Romantic partners also showed greater turn-taking behaviours during their conversations and more sustained patterns of vocal synchrony than found in conversations exchanged between strangers [94]. Because certain vocalizations and intonations are directly associated with emotional and physiological changes, speech patterns in conversations of romantic couples may reflect co-regulation, which may contribute to vocal synchrony [99]. In fact, spouses in chronically distressed marriages had synchronized the pitch and energy of their voices at higher levels when they held positive attitudes about their discussions as opposed to negative attitudes [106]. These investigations further our understanding of what features that distinguish the vocal modulations of couples conversing with one another can be discerned by listeners.

The acoustic features of a couple's speech to each other predict—and may influence—relationship outcomes, as well. Nasir et al. [107] recorded hundreds of conversations from over 100 couples taken during marriage therapy sessions over 2 years and examined their acoustic features. Some of the acoustic parameters measured included f_o, intensity, jitter, shimmer and other aspects of vocal patterns that can indicate moments of high emotion. They also had tracked the couples' marital status for 5 years and found that it was possible to predict with nearly 78% accuracy whether spouses would have an improved or worsened relationship based on the tone of voice used when speaking to their partner. In fact, the algorithm that they created from the vocal samples was better at predicting marital success of couples with serious marital issues than behavioural descriptors of the therapy sessions provided by relationship experts.

This research team followed up by investigating whether the acoustics of spoken interactions of clinically distressed spouses could provide information about therapy outcome assessments [108]. Acoustical characteristics during conversations, such as vocal intonation and intensity, were analysed both independently and in relation to each spouse and were used as cues for predicting therapy outcomes. Predictions from vocal acoustics were found to be comparable or superior to those obtained from behavioural coding by experts. Collectively, these studies point to how studying voice modulations derived from couples’ conversations can relay information about the status of a romantic relationship.

5. Conclusion

In the 1997 film ‘Contact,’ scientists detect an extraterrestrial signal containing 63 000 pages of undecipherable data originating from the star system Vega and later discover that embedded within this signal is a second, hidden message necessary for decoding the data. In recent decades, it has become increasingly evident that the acoustic features of our voices similarly comprise a second message embedded within our spoken communication, a message that can add to, modulate, or even negate the literal meaning of our speech [8,9,86]. We are beginning to decode this message, discovering that voice conveys information about a speaker's traits, such as age, sex, dominance and physical condition (e.g. [74,109–112]), and a speaker's state, including interest, affiliation, fear, apprehension or ambivalence. Further, voice can be modulated in strategic ways across social contexts to convey certain messages to the receiver.

To take one example, people seem to raise f_o and f_o − s.d. to signal affiliation, interest and lack of threat, perhaps because of acoustic associations with small size and emotional activation. Such vocal modulation is present in ‘motherese’ and ‘loverese’ and contrasts with the lowering of these parameters to signal dominance and relatively high social status. Thus, some research finds that people tend to raise f_o and f_o − s.d. under certain mating contexts, possibly to indicate interest. The seemingly paradoxical finding that people sometimes lower these parameters when speaking to particularly high-value mates may make adaptive sense if these changes are designed to feign lack of interest in highly desirable partners who may be intrigued by potential mates who appear uninterested. If so, then women who are very high in mate value may be less likely to employ this tactic, as potential mates will more often be comparatively lower in mate quality than they are. This raises the broader question of whether traits related to the speaker's relative mate value interact with states related to romantic interest in the same way that relative social status interacts with engagement in intrasexual competition to predict vocal modulation [54,75].

The studies that we have considered here point to exciting and promising areas for future research. For example, understanding how voices are modulated within and across social contexts is likely to be aided by clearer insight into how acoustic parameters are physically linked to such interpersonal variables as size, strength and health and how those parameters are perceived. From the receiver standpoint, vocal modulation during mating and competitive contexts appears to be readily discernable, and listeners prefer voice samples directed towards attractive individuals and mates. The ability to detect romantic interest through voice may be adaptive not only for identifying interested potential mates, but also in the assessment of relationship status and maintenance. Vocal shifts are also effective in influencing perceptions of speakers' rank and formidability critical for intrasexual competition. However, more work is needed to address how receiver psychology (i.e. what a receiver finds easy to detect, discriminate and/or remember) influences the design of human vocal signals (see [20]).

In addition to speech, there is recent evidence that other forms of vocalization are important in human communication, and future research could consider how these vocalizations are employed in mating and competitive contexts. Such examples include grunts [71], roars [42,113], screams [114], adult cries [115], laughter [116] and singing [117]. Additionally, some copulatory vocalizations are also under conscious control, and women in particular may alter their sexual groaning to manipulate male behaviour to their advantage [118]. For instance, female copulatory vocalizations (whether real or fake) could promote male self-esteem, which could strengthen their pair bond [118,119] or, as in other primates, could aid in the timing of ejaculation to augment conception (see [120, p. 126]; reviewed in [121]). In bonobos, it has been shown that the speed and intensity of thrusting during copulation varied directly as a function of the female partner's specific vocalizations [122].

It is well-documented that different cultures use tone of voice differently in their respective languages (see [123, pp. 182–195]); however, with notable exceptions (e.g. [117]), little cross-cultural research has explored differences in voice modulation across languages under mating and intrasexual competition contexts. Evidence that judges from different cultures, speaking different languages, can recognize vocally expressed emotion with better-than-chance accuracy suggests that the voice modulation intended for emotional expression may be at least partly driven by universal psychobiological mechanisms [124]. This may also be the case under conditions of mating and competition, but more studies comparing cross-cultural and cross-language data are needed.

As noted above, previous studies have found that speaker characteristics, such as self-perceived formidability and prestige, and motivational states, such as interest in a potential mate, influence how speakers modulate their voices across contexts. Future studies should continue to explore interactions between context and speaker characteristics, including self-perceived attractiveness relative to a potential mate, relative interest and other motivational states. Acoustic parameters are also likely to interact, both in their responses to social context and in their effects on perceptions.

More studies on these topics should be conducted across naturalistic human interactions in contexts related to human mating, such as dating experiments in the laboratory and speed-dating events. Better yet would be examining recordings from unstaged interactions at social gatherings and other venues in which mate choice and mating competition occur, although it may be difficult to obtain high-quality recordings in such settings. Recordings of relevant social interactions from across a variety of human societies will be particularly important for studying how and why patterns of vocal modulation vary. Ultimately, as these patterns across contexts begin to resolve, we will gain a better understanding of the reaction norms to be explained by evolutionary thinking, and this will lead to new hypotheses and tests of their predictions, and gradual decoding of the important messages embedded in our voices.

Data accessibility

This article has no additional data.

Authors' contributions

S.M.H. drafted the manuscript, added theoretical concepts and literature review and critically revised the manuscript. D.A.P. also drafted the manuscript, added theoretical concepts and literature review and critically revised the manuscript. Both authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.