Accent, Ethnicity, and the Politics of Persuasive Speech

Abstract

Linguistic factors are known to influence speaker credibility judgements. Past research has shown, for instance, that foreign-accented, non-native speakers are perceived as less credible than native speakers. In an influential study, Lev-Ari & Keysar (2010) presented listeners with obscure trivia facts (e.g., ‘A giraffe can go without water longer than a camel can’) produced by native- and foreign-accented speakers and asked them to indicate the truthfulness of each statement on a continuous scale. Results showed participants to perceive statements as less truthful when produced by foreign-accented vs. native speakers. However, studies seeking to replicate this effect have yielded mixed results: while some have been effective (e.g., Hanzlíková & Skarnitzl 2017, Boduch-Grabka & Lev-Ari 2021), others have shown no difference in perceived credibility between native and non-native speakers (e.g., Souza & Markman 2013, Stocker 2017, Wetzel et al. 2021). One reason for this discrepancy may lie in how credibility has been operationalized in past studies. Statements in these experiments were based on falsifiable claims regardless of the fact that, in everyday communication, judgements of speaker credibility often extend beyond factual considerations. Indeed, non-falsifiable statements – such as opinions and value judgements – are arguably more reflective of the types of information that shape perceptions of credibility in real-world communication (e.g., Friesen et al. 2015, Gampa et al. 2019). It has been shown that listeners evaluate the credibility of speakers based not only on the content of their message, but also on their perceptions of the speaker, as well as how well statements align with their own views, beliefs, and attitudes (see Mbennah & Schutte 2000). This is particularly true in highly divisive contexts, such as political discourse, where speaker credibility is often assessed based on alignment between speakers’ messages and listeners’ personal beliefs rather than on the objective truth value of a statement (e.g., Taber & Lodge 2006, Leeper & Slothuus 2014). Nonetheless, past research on the effects of linguistic factors on speaker credibility has yet to account for such realities. To address this limitation, we utilize a more nuanced and generalizable measure of speaker credibility in examining listeners’ agreement with political statements. Similar to studies on speaker credibility judgements, individuals’ agreement with political statements has also been shown to vary depending on factors such as the party or the individual to whom a statement is attributed, even when the content of a statement is held constant (e.g., Cohen 2003, Bos et al. 2013, Neumann et al. 2020, Jacobs & Pilet 2024, Jankowski et al. 2024). However, studies have yet to consider how macro-level social factors can affect listeners’ agreement with political statements. We examine this topic as it related to persuasion, which can be defined as “a successful intentional effort at influencing another’s mental state through communication in a circumstance in which the persuadee has some measure of freedom” (O’Keefe 2016: 4). Although the literature on the persuasiveness of political messaging is extensive, most studies have overlooked the role that speaker identity can have on persuasion in the political domain (Druckman 2022). This study therefore investigates the question of how native- vs. non-native speech and visual displays of ethnic background affect political persuasion. With regard to accent, we are interested not only in how native vs. non-native speech, but also whether British individuals’ agreement with left- vs. right-wing political statements changes depending on the prestige of a foreign-accented speaker’s native language. Research has shown that foreign accents are evaluated along a hierarchical scale (Dragojevic & Goatley-Soan 2022), with speakers from Western European nations typically evaluated more favorably by Western listeners than speakers from non-Western nations (e.g., Gluszek & Dovidio 2010; Lippi-Green 2012). Despite this, past research on the credibility of non-native speaker has tended to treat foreign-accentedness as a monolithic construct without accounting for how accent prestige may affect credibility judgements. One novel contribution of this study thus lies in our comparison of listeners’ agreement with speakers of Standard Southern British English (SSBE; a prestigious native accent) to speakers of two non-native accents: German-accented English (GE), a prestigious foreign accent, and Indian-accented English (IE), a relatively non-prestigious foreign accent (e.g. Coupland & Bishop 2007, Dragojevic & Goatley-Soan 2022, Sharma et al. 2022). A second contribution of this research lies in its investigation of whether audio-visual cues about a speaker can shape the persuasiveness of this speaker’s message. Research has shown that social information about speakers – such as their ethnicity – can influence how their speech in perceived. For instance, a number of studies have shown that listeners perceive the same speakers more accurately and positively when their speech is paired with a white vs. a non-white face (Rubin 1992, Kang & Rubin 2009, Babel & Russell 2015, Hanulíková 2021, Kutlu et al. 2022). Other studies have shown that congruency between audio and visual primes can result in more accurate perception of foreign-accented speech (McGowan 2015), and that mismatches between perceived ethnicity and accent can in turn lead to more negative ratings of a speaker (Hansen et al. 2017). However, because previous studies have exclusively used static photos to prime the perceived ethnicity of speakers, they are limited in the extent to which they generalize to real-life contexts in which linguistic information is most often processed simultaneously with dynamic visual information about a speaker. To counteract this disadvantage of previous research, we use videos of high-fidelity virtual avatars – with lip and facial movements synchronized realistically to speech – to cue social information about the ethnic background of our speakers in a naturalistic and representative way. Specifically, we use avatars designed to cue three different ethnic backgrounds: White British, African, and South East Asian. Pairing different ethnic primes with various accents allows us to examine how perceived ethnicity and accent interact in shaping how persuasive a speaker is. Considering factors such as speaker accent and ethnicity on political persuasion is important given the central role that social cues play in more general theories of persuasion. For instance, well-known dual-route models such as the elaboration likelihood model (ELM; Petty & Cacioppo 1985) posit that speakers’ messages can be processed either centrally – when listeners are motivated and attentive to the content of a message – or peripherally – when listeners are unmotivated and primarily attentive to external cues. Crucially, these two processing routes offer different predictions in terms of persuasive success. While persuasive outcomes via the central route depend largely on the alignment between a message and a listener’s values, outcomes via the peripheral route depend more on heuristic cues such as the perceived credibility of a speaker (see Jones et al. 2003). This dual-route model in turn allows predictions to be made regarding listeners’ political ideology and their susceptibility to persuasion (see Cakanlar & White 2023). Unlike listeners at the ideological extremes, who may focus primarily on the content of a message, more moderate or centrist listeners may lack a strong partisan bias, and therefore be more sensitive to how a message is delivered (i.e. peripheral cues) rather than solely to what is communicated. Therefore, an additional goal of this research is therefore to investigate the question of how peripheral cues relating to native-speaker status and ethnic background interact with political orientation to influence political persuasion.