Our voice plays a huge role in our social relationships (Cherry, 1953; Belin et al., 2004) and, therefore, in persuasive messages as public discourses (Rodero et al., 2022). How we use it is called prosody. Prosody is very important in non-verbal communication as it represents the set of features that we use when speaking. If the voice is our instrument, prosody metaphorically could be the song we play. Prosody is a linguist component with an essential role in communication, as it can influence the perception of a speaker (Elbert and Dijkstra, 2014; Varghese and Nilsen, 2020; Weinstein et al., 2020) and affect the cognitive processing of messages (Hirschberg and Pierrehumbert, 1986; Levi and Pisoni, 2007; Rodero et al., 2017). How a person speaks can be very relevant for the level of effectiveness and the listeners' cognitive and emotional response.

Prosody is composed of intonation, stress, and rhythm, among other features (Wells, 2007; Rodero et al., 2019). One of the main components is intonation. Intonation is composed of the different pitch variations (rises and falls) produced in spoken language (Tench, 2015). However, research about the influence of these tone or pitch variations in public speaking is scarce despite its importance. The studies about their impact on persuasive messages indicate that the most effective strategy is using a moderate pitch level with significant pitch variations, especially against monotonous styles (Burgoon et al., 1990; Hincks, 2004; Strangert, 2005; Yang et al., 2020). Jackob et al. (2011) found that substantial pitch changes increased persuasiveness and credibility. Rosenberg and Hirschberg (2009), Yang et al. (2020), and Niebuhr and Silber-Varod (2021) concluded that a higher and varied pitch is related to charisma. Rodero et al. (2017) demonstrated that the strategies with pitch variations in commercials increased listeners' attention and the ads were better recalled. Moreover, the messages with changes were perceived as the most effective and adequate. Recently, in public discourses, Rodero et al. (2022) showed that a moderate emphatic intonation was considered the most effective and credible, grabbed more attention, and elicited higher autonomic arousal. Conversely, other authors have not found pitch effects on persuasion (Elbert and Dijkstra, 2014), credibility (Chebat et al., 2007), or the attitude toward the message (Gélinas-Chebat et al., 1996).

The reason why pitch variations can work better from a cognitive point of view is that tone changes can more easily grab the listeners' attention and improve comprehension compared to a lack of variations or less variety, according to the Principle of Distinctive and Contrastive Coherence of Prosody (Rodero, 2015). When the spoken sequence is produced with pitch changes, there is an acoustic contrast between intonation rises and falls. Then, this sequence sounds more dynamic (Addington, 1968) and, hence, more effective (Rodero et al., 2017). But what is more important is that this auditory difference may attract the listener's attention and increase arousal, especially in high tones or rising contours (Hill and Miller, 2010). Together with this contrastive effect, these changes also act by differentiating the relevance of the content (less important in high pitch and more important in low pitch) and, consequently, beneficiating comprehension. However, although these changes are essential, an excessive number of variations could be counter-productive, as the sequence could sound exaggerated or unnatural. Therefore, theoretically, a moderate strategy with progressive pitch variations could be the best option (Rodero et al., 2022). On the contrary, if there are not many pitch changes, the sequence can sound monotonous and dull; thus, less effective and attractive (Glass, 1991; Knapp and Hall, 2007; Rodero, 2013). An unsuccessful speaker is monotonous (Strangert and Gustafson, 2008). Moreover, this lack of variation may affect arousal and attention due to the perception of a constant and repetitive sound. It is the phenomenon called “sensory adaptation” (Wark et al., 2007). Our brain becomes fatigued, as there is no acoustic contrast. Therefore, this monotony could hinder the emotional response, reducing participants' arousal, and being perceived as more negative. Based on these studies and ideas, we formulate the two first hypotheses of this research.

With prosody, gestures are important for a message's perception and information processing (King et al., 2020; Rodero et al., 2022). Inside this category, hand gestures are the spontaneous movements individuals make with their hands while talking (Wagner et al., 2014). These movements usually accompany speech (Iverson and Goldin-Meadow, 1998). According to Ekman and Friesen (1969), there are five kinds of gestures: illustrators (that convey semantic content), emblems (that represent conventional signs), regulators (that control conversational flow), emotional displays (for expressing emotions), and adaptors (to contact and touch). Most research has been devoted to analyzing illustrators (Harris, 2003). These gestures are coordinated with speech and aligned with semantic content (Jannedy and Mendoza-Denton, 2005). As this research also explores voice pitch, a feature related to speech, illustrators will be the gestures examined here.

Hand gestures, especially illustrators and regulators, aid lexical and semantic processing (Krauss et al., 1991) and provide some information to the listener (Kendon, 1981). Therefore, its use has shown to be beneficial for perception and cognitive processing, especially compared to the lack of them (Ekman et al., 1980; Loehr, 2004). For perception, people that use hand gestures are considered as more effective (Beattie and Shovelton, 2005), persuasive (Jackob et al., 2011; Peters and Hoetjes, 2017), credible (Maricchiolo et al., 2009), dominant (Gnisci and Pace, 2014), extrovert (Neff et al., 2010), sociable (Burgoon et al., 1990), and honest (King et al., 2020). Therefore, there is a positive perception when someone uses hand movements. For cognitive processing, some studies have shown that these gestures can improve understanding and comprehension (Graham and Argyle, 1975; Riseborough, 1981; Hostetter, 2011; Kelly et al., 2015; Dargue et al., 2019), attention (Berger and Popelka, 1971), memory (Ianı̀ and Bucciarelli, 2017; Clough and Duff, 2020), stimulate emotional reactions (Jackob et al., 2011), and clarify the discourse content (Holle and Gunter, 2007). According to the integrated system hypothesis (Kelly et al., 2010), gestures are part of an integrated system together with speech. They illustrate and draw what we are saying; therefore, they can help to understand the message.

Most research has compared the use against not use of hand gestures. For instance, Maricchiolo et al. (2009) showed that effectiveness was higher in the gesture presence condition than in the no-gesture condition. However, studies about the way to use them are scarce. Specifically, the number or intensity of gestures can be the determinant variables that affect perception and emotional response, as they can modify the expressivity level. In this regard, some studies have demonstrated positive results for a high number of gestures. Gnisci and Pace (2014) found that some speakers were considered more dominant when making a high rate of gestures. Along with this, a study of the well-known TED talks concluded that hand gestures made the speaker seem more compelling. In the most popular talks, the speakers used 465 hand gestures, whereas in the least popular, the speakers used 272 gestures (Van Edwards, 2017). However, these were self-report and perception studies that did not measure the emotional response and did not use psychophysiological techniques. The most similar study to this research was conducted recently by Rodero et al. (2022). These authors compared three different amounts of gestures. They concluded that using a moderate quantity of gestures was perceived as the most effective and credible, and a high quantity of gestures increased the participants' attention and body activation. But, along with the number of gestures, the intensity of these hand movements could also be an influent factor, as this study analyzes. A person moving both hands simultaneously could be considered more expressive than an individual moving gently only one hand when speaking. However, as with voice pitch, an excessive intensity in making gestures might be exaggerated and provoke distractions, especially as these features are visually registered. Therefore, this research hypothesizes that the medium intensity of gestures (not smooth but not strong) could be the best strategy. Strongly marked gestures could distract the audience, whereas smooth gestures could not elicit a significant emotional response. In consequence, we posit the last hypotheses of this study.

A total of forty-eight videos were recorded for this experiment, 24 on each speech, 12 by each of the four speakers. The discourses were prepared specifically for this study to control all the variables. The content was neutral and familiar for college students. The discourses were short as elevator pitches with an average of 25 s. Actors recorded both speeches with a different combination of non-verbal features: three pitch variations (few–medium–many) and three intensities of gestures (smooth–medium–strong).

Voice pitch was analyzed using Praat's acoustic analysis software (Boersma and Weenink, 2022) by measuring pitch level and pitch range. There were three different types of pitch variations (few–medium–many). The pitch values were calculated in semitones (re 1Hz). The pitch level average in the strategy with few pitch variations was 80.68 st. In the strategy with medium variations, the pitch level average in actors was 85.23 st. Finally, in many variations, the level was 90.82 st. There were significant differences among the three pitch variations (F (2,9) = 13.98, p = 0.002). The pitch range or the variation between the maximum and the minimum pitch was 19.85 st. in the strategy with few pitch variations, 24.13 st. in medium variations, and 29.33 st. in the many variations style. There were significant differences among the three pitch variations (F (2,9) = 6.85, p = 0.016). Figure 1 shows the curves of intonation with the same text.

The intensity of gestures was controlled and measured with the software Codimg, a performance analysis application that classifies gestures. The gestures used in this study were illustrators (Ekman and Friesen, 1969). The movement was slight in the smooth strategy with one hand in the low part of the body, more intense with one hand in the high part of the body for the medium strategy, and more marked using two hands chest-high for strong gestures, as shown in Figure 2. Accordingly, facial expressions were growing in intensity from smooth to strong.

The discourses were recorded in a TV studio equipped with a teleprompter for the actors to read the speeches, similar lights and acoustic conditions, a professional lavalier microphone, and a blue background. In postproduction, all the audios were normalized by using Pro Tools.

A total of one hundred and twenty university students (79 women and 41 men) aged between 20 and 30 years (M = 21, SD = 1.08) formed the sample of this study. Participants were communication students. They were divided into two groups. A total of sixty watched the first discourse (24 videos) and 60 the second one (24 videos). The 120 participants were exposed to all the experimental conditions except the type of discourse: 3 voice pitch variations (few–medium–many) and 3 intensity gestures (smooth–medium–strong) by 4 actors.

For effectiveness, we partially used a scale based on the Rodero et al. (2013) effectiveness index. The scale measured how natural, clear, and understandable the speech was. These three dimensions were measured on a 5-point scale, with 1 representing the minimum value and five the maximum. The scale had a high Cronbach's alpha coefficient (α = 0.83).

Participants rated how competent, persuasive, and pleasant the speech was for attractiveness. These three dimensions were measured on a 5-point scale, with 1 representing the minimum value and five the maximum. The scale had a high Cronbach's alpha coefficient (α = 0.90).

The emotional response was analyzed using their two main dimensions: arousal and valence (Bradley and Lang, 1994), both in self-report and physiological (arousal) and facial recognition (valence) response. The first dimension, arousal, is responsible for body activation and is the state of being awake and alert (Dawson et al., 2000). This state is produced as sympathetic nervous system response. If the level of arousal is high, there is a strong emotional activation. The second dimension, valence, determines whether this emotion is pleasant or attractive (positive) or unpleasant or avert (negative). When individuals are exposed to a stimulus, they can react positively (approaching) by activating the appetitive system or, on the contrary, with rejection by activating the aversive system (Lang and Bradley, 2013). For example, happiness is a positive emotion, while sadness is a negative one.

To measure arousal and valence in self-report, the Self-Assessment Manikin (SAM) scale was used (Bradley and Lang, 1994). Self-report arousal, defined as the intensity of the emotion, was a 5-point scale. Self-report valence, how positive and negative the emotion is, was registered on two 5-point scales: one for positive and one for negative. The final valence was computed by subtracting the positivity score from the negativity score.

Physiological arousal was measured in this study by registering the electrodermal activity. This technique measures the changes in the skin's electrical properties in response to the activity of the eccrine glands. If individuals are developing a complex task or being nervous, the sympathetic nervous system activates the body, increasing sweat secretion. The electrodermal activity registers the changes in the skin conductance provoked by sweating by applying a constant voltage. If the individuals are aroused, they are better electrical conductors, and the response is higher. Therefore, electrodermal activity is considered a measure of the sympathetic nervous system's activity to register the motivational response and increase emotional activation or arousal (Potter and Bolls, 2012). In this study, the electrodermal activity (EDA) was quantified by measuring the conductance of a constant voltage (0.5V) delivered to two 8-mm AG/AGCL electrodes placed on the participants' fingers of the non-dominant hand and connected to a Biopac bioamplifier MP-160. The signal was recorded with a sampling rate of 1,000 Hz and low pass filters.

Apart from self-report, valence can be measured by different means, such as electromyography or software for facial gesture recognition. The emotional valence is analyzed in this study using Facereader, a software for facial gesture recognition. This software measures facial movements to identify emotional expressions based on Ekman's face studies. The analysis indicates the emotional valence (positive or negative), and the kind of emotion, among other parameters.

Participants arrived at the laboratory, and they read and signed the consent form. Then, the electrodermal activity sensors were attached to the participant's fingers, and they were sitting in a comfortable armchair. Participants used headphones to listen to the videos. The order was random. After watching each video, participants had to rate the effectiveness and attractiveness levels. This procedure was repeated with all the videos. When participants finished, the sensors were removed, and they were dismissed. The experiment was conducted in two different sessions of 40 min (speech one and speech two). Students received two credits for their participation.

To test H1 and H3, a 3-pitch variation (few–medium–many) by 3 intensities of gestures (smooth–medium–strong) by 2 scripts (discourse one and discourse two) factorial MANOVA was performed on the effectiveness and attractiveness variables. There were no significant differences in scripts in any of them. The combination resulted in significant main effects for voice pitch in effectiveness, F _{(2, 118)} = 1117, p < 0.001, partial η² = 0.139. The medium style was the most effective, followed by many and few variations. The post hoc tests (Tukey) showed significant differences among all the conditions. Also, there were significant differences for attractiveness, F _{(2, 118)} = 1682, p < 0.001, partial η² = 0.195. The most attractive style was the strategy with medium variations, followed by many and few changes. The significant differences were among all the conditions, according to the post hoc tests.

Regarding gestures, the combined dependent variables resulted in significant main effects for effectiveness, F _{(2, 118)} = 18.99, p < 0.001, partial η² = 0.003. The medium intensity of gestures was the most effective style, followed by strong and smooth gestures. The post hoc tests showed significant differences among all the conditions. There also were significant differences for attractiveness, F _{(2, 118)} = 4.63, p = 0.010, partial η² = 0.001. The most attractive style was medium gestures, followed by strong and smooth. There were no significant differences between medium and strong intensities of gestures, according to the post hoc tests.

The interaction between voice pitch and gestures also was significant in effectiveness, F _{(4, 116)} = 17.62, p < 0.001, partial η² = 0.005. The medium variations with medium intensity of gestures were the most effective style, followed by medium pitch variations with strong and smooth gestures. The second group was formed by the strategies with medium pitch variations and the combination of medium, smooth, and strong gestures, in that order. Finally, the fewer effective styles were composed of the strategies with few pitch variations combined with strong, medium, and smooth intensities of gestures, in that order.

For attractiveness, the interaction also was significant, F _{(4, 116)} = 27.43, p < 0.001, partial η² = 0.008. The most attractive strategies, as in effectiveness, were the styles with medium pitch variations combined, in this order, with strong, medium, and smooth intensities of gestures. The next styles were with many pitch variations and, in this order, medium, strong, and smooth gestures. Finally, the less attractive were the strategies with few pitch variations and, in this order, strong, smooth, and medium intensities of gestures. Table 1 shows the descriptive statistics for these two variables. These results confirm H1 and H3.

To test H2 and H4, a 3-pitch variation (few–medium–many) by 3 intensities of gestures (smooth–medium–strong) by 2 scripts (discourse one and discourse two) factorial MANOVA was performed on self-reported arousal and valence. As in the other variables, there were no significant differences by script.

In self-report, voice pitch had a significant main effect on arousal, F _{(2, 118)} = 1,507, p < 0.001, partial η² = 0.178. The medium strategy elicited the highest arousal, followed by the many and the few pitch variations. The post hoc tests indicated significant differences among all the conditions. Concerning valence, there were significant differences for voice pitch, F _{(2, 118)} = 2,273, p < 0.001, partial η² = 0.246. The strategy with many pitch variations was perceived as the most positive, followed by the medium. The strategy with few changes was considered negative. The post hoc tests showed significant differences among all the conditions.

The intensity of gestures had a significant main effect on arousal, F _{(2, 118)} = 4.20, p = 0.015, partial η² = 0.001. The medium intensity of gestures achieved the highest arousal, followed by strong and smooth gestures. The post hoc tests showed no significant differences between the medium and strong gestures. In valence, there also were significant differences, F _{(2, 118)} = 3.23, p = 0.019, partial η² = 0.000. The medium intensity of gestures was perceived as less negative, followed by strong and smooth gestures. There were no significant differences between the medium and strong styles, as shown in the post hoc tests.

The interaction between voice pitch and gestures also was significant in arousal, F _{(2, 118)} = 24.74, p < 0.001, partial η² = 0.007, and valence, F _{(2, 118)} = 18.81, p < 0.001, partial η² = 0.005. The stronger arousal was stimulated by combining the strategies with medium pitch variations and strong gestures, followed by medium and smooth. The second group was composed of the styles with many pitch variations combined with medium, smooth, and strong gestures, in that order. The third group was a few pitch variations combined, in this order, with strong, smooth, and medium intensities of gestures. The most positive valence was elicited by many pitch variations combined with smooth intensity of gestures followed by many pitch variations-medium intensity of gestures, medium-strong, medium-medium, many-strong, medium-smooth. The most negative strategies were a few pitch variations combined with, in this order, smooth, strong, and medium intensities. Table 2 shows the results.

Physiological arousal was analyzed using a 3-pitch variation (few–medium–many) ANOVA. The results were significant, F _{(2, 118)} = 15.43, p < 0.001, partial η² = 0.012. The medium style achieved the highest arousal (M = 0.43; SD = 0.97), followed by many (M = 0.28; SD = 0.60), and few (M = 0.23; SD = 0.54). The strategy with medium pitch variations stimulated the strongest emotional activation followed by many and few changes. There were no significant results among many and few.

About gestures, the data were analyzed using a 3-intensity (smooth–medium–strong) ANOVA. There were no significant results, F _{(2, 118)} = 2.29, p = 0.101, partial η² = 0.002. The strategy with strong gestures achieved the greater arousal (M = 0.43; SD = 0.81), followed by medium (M = 0.34; SD = 0.91), and smooth gestures (M = 0.28; SD = 0.64).

However, the interaction between voice pitch and gestures in arousal was significant, F _{(4, 116)} = 2.54, p = 0.038, partial η² = 0.004. The combination with the strongest activation was medium pitch variations-strong gestures, followed by medium variations-medium gestures, medium variations-smooth gestures, many variations-medium gestures, many variations-smooth gestures, many variations-strong gestures, few variation-medium gestures, few variations-strong gestures, and few variations-smooth gestures. Figure 3 shows the results.

Finally, in valence, there were significant results for voice pitch measured with the facial recognition of emotions (Facereader), F _{(2, 118)} = 256.75, p < 0.001, partial η² = 0.324. The most positive style was many pitch variations, followed by medium and few changes. There were no significant differences between medium and many variations, as shown in the post hoc tests. Along with this, there were significant differences in happiness, F _{(2, 118)} = 220.52, p < 0.001, partial η² = 0.292, and sadness, F _{(2, 118)} = 113.87, p < 0.001, partial η² = 0.175. The happiest strategy was with many pitch variations, followed by medium and few changes, whereas the saddest style was the opposite, in this order: few, medium, and many. Post hoc tests showed significant differences among the three styles in happiness and sadness.

The intensity of gestures also had a significant main effect on valence, F _{(2, 118)} = 97.09, p < 0.001, partial η² = 0.153. The most positive style was the medium intensity, followed by strong and smooth. There were significant differences among the three styles, as shown in the post hoc tests. Also, there were significant differences in the two main emotions, happiness F _{(2, 118)} = 9.03, p < 0.001, partial η² = 0.017, and sadness, F _{(2, 118)} = 77.94, p < 0.001, partial η² = 0.127. Participants showed more happiness with the medium intensity of gestures, followed by strong and smooth. They expressed more sadness with the style with smooth intensity of gestures followed by medium and strong. There were no significant differences between medium and strong gestures, as shown in the post hoc tests.

The interaction between voice pitch and gestures also was significant in valence, F _{(4, 116)} = 55.91, p < 0.001, partial η² = 0.173, happiness, F _{(4, 116)} = 16.94, p < 0.001, partial η² = 0.060, and sadness, F _{(4, 116)} = 19.53, p < 0.001, partial η² = 0.068. The most positive strategy was the medium followed by all the styles with many pitch variations combined with medium, strong, and smooth intensities of gestures, in that order. However, participants expressed more happiness with the strategies of many pitch variations with medium, strong, and smooth intensities of gestures, in this order. The less happy strategy combined a few pitch variations with smooth gestures. In contrast, the strategy that elicited more sadness was a few pitch variations with smooth gestures followed by medium-strong. The less sad strategy was medium pitch variations with strong gestures. These results confirm H2 and H4. Table 3 shows the descriptive statistics.