Prosodic Elements for Content Delivery in Broadcast Journalism: A Quantitative Study of Vocal Pitch

Literature Review

Aspiring students and broadcast professionals often attempt to achieve an appropriate prosodic delivery without concrete instruction. It is possible that some individuals, through inherent ability or acquired skill, may have the increased capacity to accurately and efficiently match predetermined patterns of prosody (Winter, 2003). However, there is a need for broadcast standards to be implemented by journalism instructors and industry professionals alike (Hartman, 1986). In her book, Broadcast Voice Handbook: How to Polish Your on-Air Delivery, Utterback (2000) quotes a broadcast news director as saying

It seems that our education system has forgotten the power of voice. We don’t teach people how to use and improve their voices in school. Often by the time they get through college and their first two jobs, it is almost too late. It’s great to be a good journalist but if the voice is bad the audience won’t listen. (p. 287)

Speech prosody can also improve a listener’s overall comprehension of a speaker’s intended message. And when listeners can more clearly understand the speaker, they are more likely to assign positive personal attributes to that speaker (Acosta & Ward, 2011; Rodero, 2015) and are therefore more likely to return to that speaker. It is therefore crucial that news anchors and reporters be well trained in order to be of greatest value to their stations and viewers.

One noticeable trait of broadcast journalists is their tendency to use prosodic cues to emphasize certain words and phrases. Historically, the word “emphasis” has been used interchangeably with terms such as “stress” and “accent,” depending on the author and often on the size of utterance being discussed (e.g., syllable, word, phrase). In essence, these words or phrases will stand out above the background of the other words in the utterance. Early research (e.g., Fry, 1955; Lieberman, 1960) has established that the perceptual phenomenon of “emphasis” in speech is primarily determined by the prosodic elements of speech. Among the many aspects of vocal delivery, emphasis is one of the most important tools for a broadcast journalist, affecting not only viewer attention to news but retention of news (Hutchinson, 2009), perhaps because emphasized words are processed more efficiently than words which are not (Winter, 2003). Dilley et al. (2010) found that stressed syllables reduce uncertainty about beginnings and endings of sentences, thereby conveying meaning more efficiently. Findings from Rodero (2015) indicated that when information was conveyed by broadcasters with patterns of emphasis appropriate for the linguistic content, the listeners remembered more details from the message and found the broadcasters to be more credible.

In addition to patterns of emphasis, the global pitch range of a broadcaster’s voice can impact the attention of their audience and the comprehension of their intended message. A recent study by Rodero et al. (2017) found that radio advertisements with a greater degree of pitch variability and range resulted in an increase in audience recall of the advertisement information. This pattern of prosody was also found to increase listener’s attention and their perception of the effectiveness of the advertisement.

While listeners appear to prefer certain voice characteristics and feel drawn to certain personal attributes in the voices they listen to, credibility is unarguably another important aspect to which listeners pay attention and one which is of great importance in the field of broadcast journalism. Although the relative contribution of each prosodic cue to the perception of a speaker is not well understood, it is likely that a broadcaster’s voice pitch, loudness, tempo, and rate are all important in helping establishing rapport and credibility with an audience.

Psychological research has indicated that listeners assign certain personal attributes to speakers depending on their individual voice qualities. Findings suggest that not only is politeness judged by a speaker’s prosody but other personal characteristics such as competence, confidence, truthfulness, and likability as well. An early psychological study by Apple et al. (1979) showed how listeners used pitch and rate to assign personal attributes to male speakers. High-pitched voices were perceived as belonging to men who were less truthful, less emphatic, less potent, and more nervous. Speakers with a low rate of speech were perceived as less truthful, less fluent, less persuasive, passive, and yet more potent. Zuckerman and Miyake (1993) also found voices that were lower in mean pitch and higher in pitch range were associated with more positive personality impressions (e.g., vocal attractiveness). In 2003, Chattopadhyay, Dahl, Ritchie, and Shahin conducted a study to evaluate the impact of announcer speech characteristics on consumer responses to broadcast advertising. Results revealed low-pitched voices were rated as more attractive than high-pitched voices, especially when paired with a higher speech rate. Political scientists have found voice prosody can also impact an individual’s ability to be elected to government office, with lower pitched voices being more electable and perceived by constituents as having greater integrity (Klofstad, 2016; Klofstad & Anderson, 2018; Tigue et al., 2012). However, it is important to note that if an individual’s speaking pitch extends below that of typical voicing, glottal fry can result, which has been found to be perceived negatively as less competent, less educated, less trustworthy, and less hirable (Anderson et al., 2014).

These studies are part of an area of research that has indicated that the prosody of an individual’s speech delivery can impact their perceived credibility, trustworthiness, and likability, qualities that are particularly important to viewers and listeners when choosing a news provider. While these studies describe findings regarding perceptual qualities of preferred speaking styles for broadcasting, announcing, and advertising, they do not describe the broadcaster voice in quantitative terms. Thus, there remains a great need for quantitative research investigating the prosodic patterns of speech in advertising and broadcast journalism.

It is hoped that the information gathered in this study will begin a process to aid future students of broadcast journalism in improving delivery style and becoming more effective broadcasters by allowing them to quantitatively evaluate their own speech and voice. Such research should be done “if for no other reason, than to improve the employment odds” (Cupp, 2007, p. 145) for broadcast students. The goal here is to “help individuals sound conversational, credible, and authoritative. Those are the traits broadcast news directors are seeking” (p. 146).

Considering the previous research that has examined the effect of voice pitch on listener comprehension and perceived personal attributes, it is expected that the professional broadcasters, both male and female, will exhibit mean pitch levels that are lower than what is used by the nonprofessional voice user. In addition, it is likely that broadcast users will exhibit greater range and variability in their voice pitch. To examine these theories, the following specific research questions were posed in this study:

Research Question 1: For both males and females, what are the global pitch parameters (mean, variability, and range) of professional broadcast journalists from relatively large media markets? And do these parameters differ from the general population.

Method

Broadcast Speech Samples

Digitized audio samples of 450 news broadcast segments from 90 different television broadcast professionals were downloaded from publicly available websites and saved as .wav files on a computer hard drive for further analysis. The recordings included five segments or reports produced by an equal number of female (N = 45) and male (N = 45) broadcasters. These segments included “throws” (N = 180) and “voice-overs” or “readers” (N = 270), with an equal number of each type of segment being drawn from the female and male broadcasters. Throws and readers were used because full packages had other audio sources (interviews, nat pops, etc.) that made it very difficult to isolate the reporter’s voice. Full reports or packages were also too long for the purposes of this exploratory study.

The samples were drawn from newscasts distributed to national or relatively large media markets (i.e., Chicago and Los Angeles). The newscasts were produced by national networks and/or affiliates, namely, ABC, FOX, CNN, NBC, and CBS. The time of day of the newscast was not controlled (e.g., morning vs. evening news). Using data from those broadcasters with a publicly available age (44 of the 45 females and 32 of the 45 males), the average age of the female broadcasters at the time the samples were produced was 44.1 and 51.0 years for the male broadcasters. The samples were collected from newscasts dated from 2010 to 2018. The anchors’ places of birth were among 29 different U.S. states, as well as Australia, Korea, and Taiwan.

There was no significant difference in the mean duration of audio recordings as a function of broadcast gender. Using Adobe Audition software, the audio files were trimmed and filtered to exclude any significant background noise or music before or after the broadcast speech segment. Editing of the samples did not involve any modifications (i.e., deletions) to the middle of the audio segments; thus, only congruent samples of the broadcasters’ speech were analyzed for both throws and readers.

Pitch measurements

The pitch characteristics of each audio sample were described in terms of pitch mean, variability, and range. Using Praat acoustic analysis software (Boersma & Weenink, 2009), a series of pitch values were calculated approximately every 10 ms throughout the duration of each audio segment, as shown in Figure 1. Since each individual broadcaster has a different baseline speaking pitch, it was necessary to convert all pitch variability and range values from a linear scale (i.e., Hertz) to a normalized semitone (ST) scale to enable comparisons between the audio produced by different broadcasters. The auditory system is more sensitive to pitch changes at lower frequencies; thus, as the baseline frequency increases, a greater change in hertz is required for each subsequent ST. Using a ST scale is of particular importance when comparing male and female speakers who typically have significantly different baseline pitch characteristics to their voice.

OPEN IN VIEWER

Figure 1.

Example of the extracted pitch track (blue line) using Praat acoustic software (For interpretation of the references to colours in this figure legend, refer to the online version of this article).

Results

A detailed listing of the descriptive statistics (mean and standard deviation) for pitch mean, range, and variability can be found in Table 1. Analysis of variance was used to statistically examine whether the results differed as a function of broadcaster gender and delivery type.

OPEN IN VIEWER

Table 1.

Prosodic Measures as a Function of Broadcast Gender and Sample Type.

Prosodic Measures	Broadcaster Gender	Throw		Reader
		Mean	SD	Mean	SD
Pitch mean a	Male	154.06	22.16	153.48	19.17
	Female	203.06	21.96	204.39	26.09
Pitch range b	Male	22.40	2.24	22.25	2.61
	Female	22.74	3.68	23.51	3.45
Pitch variability b	Male	4.44	0.95	4.14	0.83
	Female	3.94	0.84	4.06	0.96

^a Measured in Hertz. ^b Measured in semitones.

Pitch Mean

As expected, a statistically significant difference was found for the measure of pitch mean across broadcaster gender, F(1, 88) = 227.91, p < .001, η² = .72, with higher mean pitch values for the female broadcasters (M = 203.1 Hz) as compared to the mean pitch produced by male broadcasters (M = 154.1 Hz). These differences are illustrated in Figure 2. Although the mean speaking pitch often varies as a function of an individual speaker’s age and vocal fold characteristics, adult male speakers typically speak at a pitch of approximately 120 Hz and adult female speakers at 210–220 Hz (Baken & Orlikoff, 2000; Ferrand, 2007; Fucci & Lass, 1999). No significant differences were found between content types (throw vs. reader) for pitch mean.

OPEN IN VIEWER

Figure 2.

Pitch mean as a function of speaker gender for broadcasters and typical speakers (approximate average from previous research).

Pitch Variability

After normalizing the pitch values to STs, a statistically significant difference between female and male broadcaster’s speech samples also was found for pitch variability within each broadcast recording, F(1, 88) = 4.16, p = .04, η² = .05. As illustrated in Figure 3 in the next section, male broadcasters exhibited a higher degree of pitch variability after the linear pitch measures were converted to the perceptual ST scale. Female broadcasters’ pitch varied by an average of 4.01 STs, while the male broadcaster’s pitch varied by 4.27 STs. Although pitch variability can vary as a function of gender, age, and dialect of the speaker, most studies of the general population have found that speakers typically vary their pitch between 2.5 and 3.5 STs (e.g., Horri, 1975; Snidecor, 1951). Similar to pitch mean, no significant differences were found between content types for pitch variability.

OPEN IN VIEWER

Figure 3.

Pitch variability as a function of speaker gender for broadcasters and typical speakers (approximate average from previous research).

Pitch Range

The statistical analysis also indicated differences in delivery pitch range across broadcasters of differing genders, F(1, 88) = 17.93, p = .01. As shown in Figure 4, the female broadcasters used a greater range of pitch (23.21 STs) than the male broadcasters (22.31 STs). Depending on what normative study you use as a comparison, female speakers in the general population tend to have pitch range values of 17–21 STs, and male speakers between 16 and 20 STs (e.g., Baken & Orlikoff, 2000; McGlone & Hollien, 1963; Snidecor, 1951). No significant differences in pitch range were found as a function of content type.

OPEN IN VIEWER

Figure 4.

Pitch range as a function of speaker gender for broadcasters and typical speakers (approximate average from previous research).

Discussion

The first research question of this study describes the global pitch parameters used by professional broadcast journalists from relatively large media markets and examines if these values differ from the general population. Interestingly, the mean pitch values used by the broadcasters in this study were different than the pitch ranges within the general population. In contrast to the theory put forth at the beginning of this study, the male broadcasters exhibited mean pitch values that were on average 24 Hz (approximately 29%) higher than what is found in the general population. This may be due to the fact that a broadcaster isn’t a “typical” voice user, in that they likely use more of the range of their pitch register, which is reflected in the increased pitch variability measures discussed below. The fact that broadcasters often deliver content by reading a copy or script also may cause their mean pitch to be elevated slightly. Previous studies have found that speakers typically produce read speech with slightly elevated pitch levels, 4–6 Hz increase for male speakers and 8 Hz for female speakers (Hollien et al., 1997). However, these differences between read and spontaneous speech would only explain a small portion of the mean pitch increase produced by the male broadcasters.

Considering the reported differences between read and spontaneous speech in the general population, the female broadcaster’s mean pitch was 7–17 Hz (approximately 4%–8%) lower than typical female speakers. Similar to the male broadcasters, female broadcasters may be using their vocal range to a greater extent than the general population, with more focus on the lower frequencies within their register. The relatively lower mean pitch values for the female broadcasters also may be due to the type of “female” voice, meaning slightly lower, that is hired more often in the field of broadcasting. The female broadcasters who work at these networks and affiliates may have anatomical vocal tract characteristics that differ from those typically found in the general female population. There is a limited amount of research that may point to why there may be a hiring bias toward female broadcasters who have a lower fundamental voice pitch.

Early research in how personality traits are associated to voice characteristics have found that voices with a lower mean pitch are found to be more truthful and trustworthy as well as more competent and persuasive (Apple et al., 1979; Klofstad, 2016; Klofstad & Anderson, 2018; Tigue et al., 2012). Findings by Zuckerman and Miyake (1993) found that personality impressions were more positive for voices that were lower in pitch. A more recent broadcast advertising study by Chattopadhyay et al. (2003) found voices with a lower pitch were judged by consumers as being more “attractive” and credible, resulting in more favorable and positive brand attitudes.

The variability ranges for the broadcasters were above or toward the upper values found in the typical population. This measure of pitch variability may be capturing larger sweeping pitch contours within a broadcaster’s utterances, as are sometimes found at the end of a sentence (pitch declination) or a question (interrogative intonation). Or the variability may be the result of more frequent smaller changes or ornamentations in a broadcaster’s delivery. Similar to pitch variability, the values for overall pitch range were somewhat higher than the pitch range values found for the general population. These results are not surprising considering that as professional voice users broadcasters would be expected to use their pitch range more extensively when delivering a news story.

The second research question examined if global pitch parameters used by professional broadcasters differ as a function of gender. The results of this study indicate several gender-related differences between how the female and male broadcasters used their voice to deliver the news. It is interesting that the female broadcasters’ mean pitch was found to be lower than the pitch levels found in female speakers within the general population; however, the male speakers’ mean pitch was somewhat higher than what is found in the general population. One explanation for this pattern may be a societal or hiring bias toward female broadcasters with a lower overall pitch range. Justifiable or not, female speakers with lower fundamental pitches in their vocal delivery have been shown to be more credible than those with higher, younger sounding voices (Chattopadhyay et al., 2003; Klofstad, 2016; Tigue et al., 2012). Although the pitch values for the female broadcasters were lower than the general population, their pitch did not extend into pitch ranges that would elicit a “glottal fry” sound quality to their voice. Glottal fry occurs when a speaker lowers the pitch of their voice by relaxing the vocal folds to such a degree that the vocal folds start to vibrate with an atypical pattern. Research has indicated that female speakers’ voices with significant amounts of vocal fry are found by listeners to be less competent and less trustworthy than typical voicing patterns (Anderson et al., 2014).

Similar to their male colleagues, female broadcasters tended to use more pitch variability and pitch range than typical speakers. The male broadcasters exhibited a higher degree of variability in their voice, whereas the female broadcasters used a greater range in pitch. Once again it is important to note that these measures already have been perceptually normalized into STs to account for differences in baseline pitch levels across the gender of the broadcasters.

Finally, the third research question studied if the broadcaster’s pitch parameters differed as a function of the linguistic story form or type. It was interesting that the prosodic analyses found no significant differences in a broadcaster’s vocal mean pitch, variability, or range as a function of the story form being delivered—throw, voice-over, or reader. Broadcasters don’t seem to switch speaking styles when going from one basic story form to another.

Conclusions

Findings indicated that compared to typical speakers in the general population, male broadcasters speak with an elevated mean pitch, more pitch variability, and use more of the range of their pitch register. The increase in mean pitch is likely due to using the upper part of their voice range more extensively than typical speakers. Unlike the stereotype we sometimes see in TV shows and comedic films where the anchor has a “deep” or low-pitched voice, the findings of the study indicate that many male broadcasters exhibit a typical voice pitch during their delivery—but one that is somewhat expanded in range and intrinsic variability. For new broadcasters, the important factor is to establish a speaking pitch that is natural to the speaker’s own voice and use their pitch range in a somewhat more dynamic manner.

Unlike their male colleagues, the findings of this study indicate that female broadcasters tend to deliver the news with mean pitch values that are lower than female speakers in the general population. The reason for these differences is beyond the scope of this study; however, factors such as hiring bias and the effect of voice pitch on listeners’ perception of personality attributes may impact the voice delivery of female broadcasters. Considering that societal norms often change over time, more recent research is needed that further examines the association between voice characteristics and impressions of personality.

Despite the exploratory nature of this study, it is hoped that these findings can provide some guidance to students of broadcasting regarding how successful broadcasters use their voice in terms of speaking pitch. Professional broadcasters seem to use the range and variability of the voice more than the general population, but not in an extreme manner. Although we don’t have data regarding the conversational speech characteristics of the broadcasters included in this study, the nature of their speech when broadcasting does not seem to differ drastically from the values that might be found in their everyday conversations.

We also hope this initial study serves as a starting point for future research on prosody in broadcast journalism to help establish clearer guidelines and coaching techniques. We believe the next step is to examine, define, or control for more specific variables such as actual story content, story part, use of emphasis, lexical structure and grammar, and more.