Introduction The ability to persuade

Introduction The ability to persuade

The limits of my language mean the limits of my world.

––Ludwig Wittgenstein (1922)

The ability to persuade––to convince a stalwart listener or sway the undecided masses––

is a prized human communicative skill. Social scientific research investigating the phenomenon

has demonstrated that messages intended to persuade usually contain implied or overt arguments

supporting an advocated position (O’Keefe & Jackson, 1995). These embedded arguments

(e.g., Chaiken, 1979), or similarity (e.g., Busch & Wilson, 1976) have frequently been shown to

enhance, detract, or even dictate the outcome of a persuasive message (Wilson & Sherrell, 1993).

In a similar yet predominantly overlooked fashion, a person’s linguistic style, such as their accent

(i.e., manner of pronunciation), “appears to have a powerful influence on the judgments of

perceivers,” suggesting it also may have an impact on persuasive appeals (Foon, 1986, p. 521).

Consequently, this dissertation seeks to examine the relative influence of source accent and

argument quality on persuasion under different conditions.

Source Accent and Persuasion

Empirical literature examining the phenomenon of the social evaluations of speech (i.e.,

language attitudes) has consistently found that speakers are readily judged on various evaluative

traits based solely upon their accent (Giles & Watson, 2013). Similar to other forms of social

Firstly, a speaker’s accent may act as a cue to social categorization, leading to the

activation of relevant stereotypes, which guide subsequent judgments. Status stereotypes are

largely based on perceived socioeconomic status (Fiske et al., 2002). In contrast, solidarity

stereotypes tend to reflect ingroup loyalty (Giles & Watson, 2013). Given that people tend to

stereotypically associate most foreigners with lower socioeconomic status and categorize them as

outgroup members (i.e., nationally), foreign-accented speakers are frequently judged less

favorably on both dimensions in comparison to native-accented speakers (see Fuertes et al.,

2012; for exceptions, see Stewart, Ryan, & Giles, 1985).

Secondly, a listener’s relative ease or difficulty in cognitively processing accented speech

(i.e., processing fluency) varies for different accents (Giles & Watson, 2013). This metacognitive

typically more difficult to process than native-accented speech (e.g., Cristia et al., 2012),

investigations utilize well-established theories of persuasion to guide or explain findings (for

exceptions, see Lalwani, Lwin, & Li, 2005; Morales, Scott, Yorkston, 2012). Nevertheless, on the

foundation of these limited, mostly atheoretical, and typically main effect studies (e.g., are some

accented speakers more persuasive than others?), a general finding has emerged: foreignaccented

speakers from non-Anglo regions (e.g., Latin America or Asia) are typically perceived as less

persuasive than their native-accented counterparts (e.g., DeSheilds, Kara, Kaynak, 1996; Tsalikis,

DeShields, & LaTour, 1991). The dominant explanation for this finding is that accents which

elicit more favorable social judgments tend to be considered more persuasive (e.g., Dragojevic,

Savage, Scott & McGinnis, 2018; Morales, Scott, Yorkston, 2012). Consequently, since non-

Anglo foreign-accented speakers are typically judged less favorably than nativeaccented speakers

are manipulated (e.g., DeShields et al., 1997; DeShields & Kara, 2011). It is, therefore,

reasonable to propose that the effects of source accent on persuasion may be contingent on other

variables, such as message (e.g., argument quality) and listener characteristics (e.g.,

how these variables might interact to influence persuasion.

Contemporary theories of persuasion, such as the well-established ELM (Petty &

Cacioppo, 1986b), propound two relatively distinct processing routes to attitude change: The

systematic and methodical route (i.e., central processing) and the quick and recursive route (i.e.,

peripheral processing). When receivers process a persuasive appeal via the central route, they

rely on the relevance and quality of message arguments to guide their evaluation of the appeal.

Under central processing, messages that contain strong arguments are typically more persuasive

than messages that contain weak arguments (Carpenter, 2015; O’Keefe, 2013). In contrast, when

receivers process a message via the peripheral route, they rely on basic inferential (or peripheral)

cues or heuristics (e.g., source expertise) to guide their evaluations of the appeal (Kitchen et al.,

2014). Under peripheral processing, messages that contain positively-valenced peripheral cues

receivers are focused on message content and base their judgments on argument quality rather

than source characteristics; strong arguments should be more persuasive than weak arguments,

regardless of source accent. On the other hand, when a message is processed peripherally, source

should be more persuasive than negatively evaluated sources (e.g., foreign-accented speakers),

irrespective of argument quality.

capability (ability) to elaborate, or engage in issue-relevant thinking (Petty & Cacioppo, 1986a).

When receivers’ motivation and ability to process a message are both high, receivers engage in

high levels of elaboration and process the message via the central route. Conversely, when

receivers’ motivation and/or ability to process a message are low, receivers engage in low levels

of elaboration and process the message via the peripheral route.

Various factors determine elaboration likelihood by influencing receivers’ motivation

and/or ability to process a message. One factor that has received considerable research attention

is receivers’ level of issue involvement, or the degree to which receivers perceive the message

Based on what has been described so far, the relative impact of argument quality and

source accent across high and low issue involvement conditions should be as follows. Under high

issue involvement conditions, persuasion should primarily depend on argument quality, and

source accent should have little effect. All else being equal, strong arguments should be more

persuasive than weak arguments, regardless of the source’s accent. Under low issue involvement

little effect: All else being equal, native-accented sources should be more persuasive than most

foreign-accented sources, regardless of argument quality.

between source accent, argument quality, and issue involvement on persuasive outcomes is based

on the premise that a source’s accent strictly functions in the role of a peripheral cue during the

evaluation process. However, the ELM postulates that persuasive variables can, in some

contexts, take on multiple roles (i.e., central cue, peripheral cue, biasing cue, elaboration

moderator; Booth-Butterfield & Welbourne, 2002). This dissertation argues that source accent is

one such variable.

Source characteristics can function not just as peripheral cues but also in multiple ways

depending on the context (Brinol & Petty, 2009). In particular, recent findings in language

attitudes (e.g., Dragojevic, Giles, Beck, & Tatum, 2017) and persuasion (e.g., Guyer, Fabrigar, &

Vaughan-Johnston, 2018) suggest that a source’s accent may also function as an elaboration

moderator, by influencing recipients’ ability to engage in extensive issue-relevant thinking. As

noted earlier, foreign-accented speech is typically more difficult to process than native-accented

foreign accent speech is considerable (i.e., medium to high levels), it is likely to also impede a

suggests that even under high involvement conditions, source accent may play a role because it

can influence receivers’ level of elaboration and determine how they process the message.

argument quality, and issue involvement on persuasion needs to be qualified. Specifically, under

high issue involvement conditions, strong arguments should be more persuasive than low-quality

arguments when the source’s accent is native. When the source’s accent is foreign, a receiver’s

ability to process the message is likely to be diminished, increasing the probability that they

engage in peripheral processing. Subsequently, if a receiver is peripherally processing, the

quality of message arguments should have little, if any, effect, even under high involvement

conditions. In sum, this suggests that accent can play multiple roles in persuasion: it can function

as a peripheral cue and an elaboration moderator.

General Overview and Organization

issue involvement conditions. Utilizing the ELM as a theoretical framework to understand the

guiding fundamental processes, two focal propositions are explored.

message quality: Native-accented sources should be more persuasive than their foreign-accented

counterparts, regardless of the quality of message arguments.

Secondly, under high involvement conditions, source accent can function both as an

elaboration and lead to central processing, as long as receivers’ ability to process the message is

not diminished. When the source is a native-accented speaker, these conditions should be

satisfied and persuasion should depend on argument quality, not source accent: Strong arguments

should be more persuasive than weak arguments, regardless of the source’s accent. In contrast,

when the source has a particularly strong or unfamiliar foreign accent (e.g., a non-Anglo variety),

these conditions may necessitate that a receiver expend considerable cognitive effort in order to

process the message. Subsequently, the increased effort required will likely diminish a receiver’s

ability to elaborate, leading them to engage in peripheral processing. In this instance, the source’s

foreign accent functions initially as an elaboration moderator lowering receiver elaboration, and

then as a peripheral cue. Consequently, the persuasive outcome will once again be contingent on

the evaluations of the source’s accent; as under low elaboration, the message’s

arguments should have little, if any, effect.

Together, these main propositions suggest that certain source accents (in this case, non-

which source accent impacts persuasion warrants experimental investigation.

In order to establish context, this dissertation first provides an overview of language

of the limited empirical research detailing how language attitudes influence message

persuasiveness. It then proceeds to describe the central tenets of the ELM and explain in detail

how its two routes to information processing and attitude change provide theoretical elucidation

regarding the function and role that source accent plays in persuasion. Lastly, it proposes an

experiment designed to investigate the conditions under which source accent does and does not

impact persuasion under the ELM's specific theoretical mechanisms.

unfamiliar or dissimilar to a listener’s own, it is typically deemed harder and more effortful to

own speech (e.g., Cristia et al., 2012; Munro & Derwing, 1995a, 1995b), the more it potentially

disrupts and reduces the listener’s processing fluency (e.g., Lev-Ari & Keysar, 2010). This

disruption then leads to negative evaluations of the speaker (Dragojevic, Giles, Beck, & Tatum,

2017). In this context, there are two underlying mechanisms primarily guiding the outcome:

naïve theories and associated affect (Dragojevic, 2019).

Naïve theories refer to the various learned or acquired assumptions that guide people's

domain-specific judgments of the information they are receiving and why it is easy or difficult to

process (Alter & Oppenheimer, 2009; Dragojevic, 2019). In other words, naïve theories "bridge

the gap between the experience of fluency and its implications for a particular judgment," by

imposing inference, such as a speaker's assumed status and solidarity (Alter & Oppenheimer,

2009, p. 2; Shawarz, 2015). For instance, in a communication context, a receiver of a message

may have the belief that the effectiveness of the communication and their own resultant

Thus, the processing fluency of a given cognitive task is also hedonically marked: fluent

Oppenheimer, 2009; Dragojevic, 2019). This associated affect is then drawn upon to make

subsequent judgments (Shawarz, 2015). The task of processing a speaker's style of speech

functions in a similar way.

When speech processing is fluent, the associated affect it elicits is positive, favorably

biasing language attitudes. In contrast, disruptions, or difficulties in processing lead to

disfluency, eliciting negative affect, which can then negatively bias a listener’s language

attitudes. For example, disruptions to a receiver's processing fluency from external noise (e.g.,

Dragojevic & Giles 2016) or a speaker's non-native accent (e.g., Lev-Ari & Keysar 2010)

indirectly impacts language attitudes via affect (Dragojevic, 2019; Dragojevic, Giles, Beck, &

Tatum, 2017). In this way, foreign-accented speakers may be rated less favorably than

nativeaccented speakers because the former elicit a more negative affective reaction during the

cognitive task

reinforce, or change the responses of another, or others” (Stiff & Mongeau, 2016, p. 10).

Compared to other contextual veins within language attitudes, persuasive messages delivered by

variety of accents used to deliver a persuasive message can influence the overall attitude toward

the advocated position (e.g., Lalwani, Lwin, & Li, 2005; Morales, Scott, Yorkston,

2012), purchase intentions (e.g., DeSheilds, Kara, Kaynak, 1996; Tsalikis, DeShields, & LaTour,

these studies primarily investigated the main effects of whether some accented speakers were

more persuasive than others. Within these studies, a general finding has been shown to emerge.

more persuasive than appeals delivered by non-Anglo foreign-accented speakers (e.g.,

DeSheilds, Kara, & Kaynak, 1996). This is in line with the earlier discussion on social

categorization and stereotyping at the beginning of this chapter. Specifically, foreign-accented

speakers, such as those from non-Anglo or non-native English-speaking foreign countries, tend

to be less favorably evaluated than native-accented speakers from the US, even allowing for

notable exceptions such as RP speech (Morales, Scott, Yorkston, 2012).

The explanation for this general finding, therefore, expands the previous discussion––

certain accented speakers, such as native English speakers, elicit more favorable social

judgments. In turn, this upgrading in social evaluations leads them to be considered more

attributed more status tend to be more persuasive. Consequently, because non-Anglo

former tend to be less persuasive (Tsalikis, DeShields, & LaTour, 1991). Although solidarity

attributions are also positively associated with persuasiveness, their effect appears to be less

pronounced than the effect of status attributions (Dragojevic, Savage, Scott, & McGinnis, 2018).

Based on what has been described so far in this section, non-Anglo foreign-accented

speakers are typically judged less favorably than native-accented speakers along the evaluative

dimensions of status and solidarity. Due to the comparative downgrading on these evaluations––

particularly on status-based traits–– non-Anglo foreign-accented speakers are, in turn, frequently

perceived as less persuasive than their native-accented counterparts.

However, despite strong empirical support for this explanation, some studies have

displayed alternate or seemingly inconsistent findings when certain message qualities (e.g.,

Giles, Williams, Mackie, & Rosselli, 1995; Morales, Scott, Yorkston, 2012) or listener

characteristics are manipulated (e.g., DeShields, 2015; DeShields et al., 1997; DeShields &

Kara, 2011). For example, empirical research grounded in theory demonstrates that a message’s

quality (i.e., how strong or weak are its embedded arguments) largely dictates the persuasive

outcome, even more so than a message’s source (Carpenter, 2015; Wilson & Sherrell, 1993).

Research indicates that receivers who carefully examine and scrutinize the merits of the

& Cacioppo, 1986a). This careful examination by receivers typically means that appeals

containing strong quality arguments will increase a message’s perceived persuasiveness and

weaker quality arguments will decrease the appeal’s persuasiveness. However, if a receiver is

unable or unwilling to scrutinize an appeal’s information, the quality of the message matters

little. In this instance, the success of the persuasive outcome is more likely to be dictated by

factors external to its information, such as the source’s accent (Carpenter, 2015; Petty &

Cacioppo, 1986b).

The current literature has not fully explored the potential effects of source accent and

argument quality on persuasion under varying degrees of issue-relevant thinking. Consequently,

it is reasonable to propose that the impact of source accent on persuasion may be contingent on

these variables, which have been shown to interact with and influence persuasive outcomes in

other contexts (Petty & Cacioppo, 1986a; Wilson & Sherrell, 1993). Therefore, to examine the

potential impact of source accent and these variables on persuasion, it is prudent to overview the

that a source’s accent may be functioning in previously unexplored ways, the impact of source

The Elaboration Likelihood Model (ELM; Petty & Cacioppo, 1986b) is a general theory

of attitude change that offers a dual mental process framework for understanding the fundamental

processes involved in effective persuasive communication messages. The ELM posits that

“variations in the nature of persuasion” depend on the likelihood that a receiver of a persuasive

appeal is willing and/or able to engage in elaboration (i.e., issue-relevant thinking) pertinent to

the issue at hand (O’Keefe, 2008, p. 1475). Based on the extent of elaboration, which ranges on a

continuum from low to high, two relatively distinct processing routes to attitude change are

possible: the systematic and methodical route––central processing––and the quick and recursive

route––peripheral processing.

The central processing route of attitude change occurs when a receiver’s elaboration is

relatively high. This form of systematic processing involves the receiver carefully examining the

message’s true informational merit is typically sidestepped in favor of simpler decision-making

example, a receiver using peripheral processing may be persuaded by either how attractive or

credible they perceive the source (i.e., sender) of the message to be, rather than by scrutinizing

the actual arguments presented in the message. In comparison to higher elaboration levels, which

engage central processing, attitudes shaped under lower levels of elaboration are typically less

temporally persistent and less predictive of subsequent behavior (O’Keefe, 2013).

Elaboration Continuum: Influences on the Degree of Elaboration

The central and peripheral routes to persuasive information processing are considered two

distinct avenues towards attitude formation or change. However, the route a receiver takes does

not have to be solely one or the other, as the path is viewed as an elaboration likelihood

continuum (Petty & Cacioppo, 1986b). The elaboration continuum refers to the probability that a

receiver will engage in effortful thought, with more elaborate processing at one end of the

continuum and effortless and less elaborate message processing at the other. Consequently, the

relative importance of the decision-making information or procedures that typically guide the

persuasive outcome under each processing route varies depending on the extent of a receiver’s

elaboration. For example, in a hypothetical context, if a receiver were at the lowest point on the

the capability (i.e., ability) to engage in issue-relevant thinking (O’Keefe, 2008).

Motivation to Elaborate. The willingness to engage or be motivated to evaluate a

presented appeal affects the persuasive outcome by influencing the degree of a receiver’s

elaboration. Several underlying factors have been found to impact a receiver’s degree of

Consequently, the stronger a person’s need for cognition, the more likely they are to want to

low need for cognition are typically less motivated to elaborate as they innately tend to avoid

engaging in cognitive effort, even when they are capable of doing so (Petty & Cacioppo, 1986b).

level is a receiver’s ability to process message information (Booth-Butterfield & Welbourne,

2002). When a receiver has the ability to elaborate extensively, they can closely examine the

given information, scrutinize it, and present counterarguments (O’Keefe, 2008). This extensive

engagement in issue-relevant cognition (i.e., high elaboration) leads a receiver to primarily

process information centrally, which increases the likelihood that the strength of a message’s

arguments will guide the persuasive outcome. However, various personal and outside factors can

impact a receiver’s ability to elaborate, such as their self-validation (Briol & Petty, 2009) or the

number of times a given message is repeated (Cacioppo & Petty, 1989). Three particular factors

are considered prominent in the context of persuasive verbal messages: the presence of

distraction (Petty, Wells, & Brock, 1976), the message’s level of complexity (Hafer,

claim is pro- or counter-attitudinal to the predilection of the receiver’s dominant thought

When a receiver interprets a message's content, they gauge the source’s position relative

to their own. Subsequently, if they perceive the sender’s attitudinal position in the message to be

similar to their own (i.e., pro-attitudinal), they are likely to agree with the advocated position in

the message, which then generates favorable thoughts. However, if the source's stance appears

counter to their own (i.e., counter-attitudinal), the receiver will be prone to negative or counter

argumentative thoughts about the advocated position (O'Keefe, 2008).

For instance, if a receiver engages in high levels of elaboration and a persuasive message

is counter-attitudinal and has a weak argument, a successful persuasive outcome is unlikely. The

reduction in persuasiveness occurs because the receiver is almost certainly centrally processing,

which, in this context, leads them to generate fewer positive thoughts and engage in counter

argumentation based on the message’s unfavorable information (Petty, Wells, & Brock, 1976).

persuasiveness if it then reduces the receiver’s ability to produce counter argumentative thoughts

by leading them to process the information peripherally instead of centrally (Petty, Wells, &

elaboration level, increasing the probability of peripheral processing, and thus reducing the

potential for favorable thought production elicited from the message’s convincing information.

rarely, if ever, be persuaded by counter-attitudinal messages. Subsequently, prior research has

message, with strong and cogent arguments also evoking positive thoughts (O'Keefe, 2013).

their message quality. In high elaboration conditions, a message’s argument quality takes a

commanding role, with strong or high-quality arguments (e.g., logical, verifiable facts) leading to

more successful persuasive outcomes than weak or low-quality arguments (e.g., illogical,

specious claims) (Carpenter, 2015). Under high elaboration conditions, message quality

predominantly impacts the persuasive outcome because the message’s content is subjected to

increased levels of scrutiny. Therefore, message arguments that are perceived as robust and with

sound reasoning will lead to more positive issue relevant thoughts. In contrast, message

arguments perceived as illogical or based on specious evidence will lead to more negative issue

relevant thoughts (Petty & Cacioppo, 1979).

accents in varying persuasive settings (evaluations towards the product, public service

message quality (i.e., PSA argument weak or strong) in order to examine the extent to which

each impacted persuasion. Findings showed that recipients were predominantly persuaded by

strong rather than weak arguments. In other words, strong arguments were still given more

weight than weak arguments, regardless of the speaker’s accent, presumably because participants

were processing the message centrally.

At this stage, the discussed findings and the ELM predominantly suggest that a source’s

accent should primarily function as a peripheral cue. A source’s accent should then substantially

impact the persuasive outcome when a receiver’s elaboration level is low. Subsequently, under

these peripheral route conditions, an appeal’s argument quality should have little impact in

guiding decisions. In contrast, message quality has its most substantial impact on the persuasive

outcome when a receiver’s elaboration is high, with source accent having little impact under a

receiver’s central route processing. This description would be the case if a source’s accent only

centrally. However, participants were told to “focus attention explicitly on the persuasiveness of the

advertising message” (p. 40). It is, therefore, plausible to assume that the recipients were displaying more

extensive levels of elaboration.

functioned as a peripheral cue. While some variables like argument quality predominantly

function as central cues (i.e., issue-relevant argument), source characteristics, depending on the

context, have been shown to function in additional ways (Brinol & Petty, 2009).

Source Factors in the ELM

As noted previously, source factors are an extrinsic aspect of the communication situation

and thus typically function as peripheral cues. However, according to the ELM, source factors

can function in multiple roles based on different conditions (Petty & Cacioppo, 1986b). Under

Source Factors as Peripheral Cues. As noted earlier, source factors (e.g., credibility,

attractiveness, and likability) can be peripheral cues. Source factors as peripheral cues typically

function as simple acceptance or rejection cues, impacting the persuasive outcome when a

receiver is peripherally processing (Petty & Cacioppo, 1986a). The lower a receiver’s elaboration

A source’s accent has also been shown to function as a peripheral cue under lower

shown to have a stronger effect on purchase intentions under low involvement conditions,

compared to high involvement conditions. In other words, source accents evaluated favorably

(unfavorably) in language attitudes (e.g., status attributions) under minimal elaboration

conditions typically solicit more (less) message persuasiveness.

Source Factors as Central Cues. When a receiver is engaged in extensive elaboration,

they are likely to be centrally processing the message. Accordingly, the information in a message,

such as the quality of its embedded arguments, primarily guides the relative persuasiveness of the

appeal (O’Keefe & Jackson, 1995; Petty & Cacioppo, 1986a). Under these elaboration

conditions, source factors typically have little to no impact on the persuasive outcome. However,

when source factors are perceived as substantive and issue-relevant pieces of evidence for the

argument, they can function as central cues (or arguments: Brinol & Petty, 2009). In

experimental conditions, a particular source factor functioning as a central cue typically only

impacts the persuasive outcome under certain contexts, such as when message quality is kept

instance, group identity has been shown to enhance persuasion under central processing

process (cf. Giles, Williams, Mackie, & Rosselli, 1995).

Source Factors as a Biasing Cue. Another way source characteristics can function is by

receiver’s mind. Subsequently, under high elaboration conditions, if message quality is

of the message. For example, suppose a message’s source is considered to have substantial

almost inevitable. For example, in the previously mentioned study by Morales, Scott, and

Yorkston (2012), which investigated the effects of source accent and message quality on

persuasion, biased processing is seemingly evident under the study’s central processing

conditions. As mentioned before, a message’s quality was more impactful than the source’s

accent in ratings of message persuasiveness. However, favorable judgments of the source’s

accent were also shown to have an evaluation halo effect, regardless of the message’s argument

quality. In other words, even though strong arguments were still given more weight than weak

arguments, favorable judgments of a given source’s accent still raised the relative persuasiveness

of their arguments.

Source Factors as Elaboration Moderators. Variables that influence persuasion by

affecting a receiver’s extent of elaboration about the given persuasive communication are

described as elaboration moderators (Booth-Butterfield & Welbourne, 2002). Source factors have

been shown to moderate elaboration by impacting a receiver’s ability to engage in issuerelevant

thinking about a message’s arguments. Source factor variables that cause a receiver to be

distracted can increase or decrease elaboration via their impact on processing ability (Brinol &

Petty, 2009). These variables can sometimes initially function as elaboration moderators, and

then act as peripheral cues if their impact as a moderator significantly reduces elaboration (Brinol

informational merit is not as crucial to the persuasive outcome.

Moreover, since rapid speech is often associated with higher competence and intelligence

hand, if an appeal is strong in quality, this reduction in the ability to elaborate reduces the success

of persuasion (Smith & Shaffer, 1995). In this instance, a receiver no longer fully comprehends

the message’s strong arguments, which, with a slightly less rapid speech rate, would lead to more

favorable thought production and persuasive success. Instead, the receiver is peripherally

processing and basing their message judgments on peripheral cues rather than the message’s

informational merits. While the above may still lead to persuasive success if the rate of speech is

associated with favorable heuristics, the overall perceived strength, temporal persistence, and

predictability of subsequent behavior will likely remain lower than if the claim was centrally

processed (O’Keefe, 2013).

Recent additions to persuasion literature suggest that the impact of source effects may be

not only impactful at the direct level of cognition discussed above, but also at second-order

cognition––or metacognition (Brinol & Petty, 2009; Petty et al., 2002). A person’s primary

thought involves the initial association of an object with an attribute or feeling (e.g., this message

Schwarz, 2015). As such, an individual’s processing fluency is not directly a form of information

associated with a given cognitive operation (Oppenheimer, 2009). Consequently, disruptions or

distractions (e.g., external noise or language complexity) to the processing of information––in

this case, evaluating the merits of a persuasive appeal––have also been shown to not only impact

primary thoughts but also impact an individuals’ secondary thoughts (i.e., via the metacognitive

experience of processing fluency). Wherever a particular task falls on the continuum, its

interpretation will lead to different judgment outcomes as processing fluency is largely

contextual. Naïve theories (i.e., the various learned or acquired assumptions), as previously

mentioned, largely govern the basis for the interpretation and the associated affect influences the

experience of the informational processing (Alter & Oppenheimer, 2009).

In most contexts, when a receiver is exposed to a task or information, if it feels easy to

process (i.e., fluent processing), several positive associations are activated (for an exception, see

Galak & Nelson, 2011). For instance, fluent processing has been shown to elicit positive

elevations of credibility, intelligence (Oppenheimer, 2006, 2009), and liking (e.g., Reber,

information (e.g., the use of jargon, or complex language etc.) have been demonstrated to reduce

message acceptance (e.g., Bullock, Colón Amill, Shulman, & Shulman, 2019) and reduce topic

the persuasive outcome, it is possible that a source’s accent functions similarly under select

contexts. Specifically, and mentioned previously in this chapter’s section on language attitudes,

accented speakers deemed as unfamiliar or dissimilar to a listener’s own are often considered to

have strong or heavy accents and their speech is thus perceived as being more difficult to process

than more familiar or similar accented speech (Cristia et al., 2012). For example, non-Anglo

foreign accented speakers have been shown to disrupt US listeners’ processing fluency to a much

higher degree than native speakers, typically also resulting in comparatively less favorable status

and solidarity judgments (Cristia et al., 2012; Dragojevic & Goatley-Soan, 2019).

Accordingly, if extensive difficulties in processing fluency function similarly to

Reynolds, & Obertynski, 1996) elaboration ability, a source’s non-Anglo foreign accent may

fluency-based and/or linguistic categorization and stereotyping evaluations, rather than the

foreign-accented speakers, most non-Anglo foreign-accented speakers will be persuasively

downgraded compared to native-accented speakers.

still, in at least two contexts, lead to relatively positive persuasive outcomes relative to a

nativeaccented source. First, referring back to what was previously discussed, if a receiver is

motivated, disruptions to a receiver’s ability to elaborate are not always sufficiently distracting

enough to divert information processing to the peripheral route (Petty, Wells, & Brock, 1976). In

fact, a curvilinear relationship can occur, in which messages with strong arguments and no

disruptions present can become slightly more persuasive if minor external disruptions do occur.

Accordingly, in this context, if a non-Anglo foreign-accented source disrupts receiver processing

to what is equivalent to only minor levels of distraction and a native-accented speaker causes

none, the foreign-accented source could, in comparison, end up being considered slightly more

persuasive. Second, a non-Anglo foreign-accented source could also improve the persuasiveness

effortful or difficult to process than native speech (Cristia et al., 2012) can disrupt listeners’

processing fluency (e.g., Lev-Ari & Keysar, 2010). Specifically, VE-accented speech disrupts

listeners’ processing fluency to a higher degree than native GAE-accented speech (Dragojevic-&

Goatley-Soan, 2020). Therefore, it is reasonable to suggest that disruptions to processing fluency

due to a VE-source’s accent may negatively impact a receiver’s ability to engage in extensive

elaboration, irrespective of their level of issue involvement. If this is indeed the case, the

outcome description of source accent and argument quality needs to be qualified under high

arguments should be more persuasive than weak arguments. In contrast, if a source has a VE

accent, the conditions necessary for central processing are unlikely to be satisfied. Notably, the

increased effort required to process a VE-accented source’s speech may disrupt a receiver’s

ability to elaborate, increasing the probability that they will engage in peripheral processing. In

this situation, a source’s VE accent functions initially as an elaboration moderator by reducing

receiver elaboration and then as a peripheral cue. Consequently, under these low elaboration

H5: Under high issue involvement conditions, strong arguments will be more persuasive

research in a domain of seeming literary dearth and addresses the call for “researchers in

language and persuasion to conduct more theoretically grounded research” (Hosman, 2002, p.

383). This chapter presented the ELM as a framework that provides a theoretically grounded

rationale for some of the seemingly inconsistent past research findings. Moreover, it also

provides a clear theoretical vehicle for investigating the various hypotheses presented at the end

of the chapter which seek to clarify the relative influence of source accent and argument quality

on persuasion under various elaboration conditions. The next chapter, Chapter Three, provides a

detailed overview of the methods used to test this dissertation’s five primary hypotheses.

CHAPTER 3. METHODS

be discussed throughout the section as their findings help verify the measures, manipulations,

accent) x 2 (argument quality: weak or strong) x 2 (involvement: low or high) factorial design.

Procedure

Participants were recruited through SONA for moderate reward (e.g., class credit), after

which they then took an online survey containing the experimental study. Following prior studies

(e.g., Petty, Cacioppo, & Goldman, 1981) participants were presented with a contextual scenario.

The contextual scenario was that the University of Kentucky is currently undergoing an

academic re-evaluation and that the university President is seeking recommendations about

policy changes to be instituted. In order to obtain a variety of opinions about the university and

its future, participants were told that the university President had asked several individuals and

groups to prepare policy statements. The participants were then further told that the policy

statement that they were about to listen to had been recorded for possible distribution and that the

College of Communication and Information was cooperating with the university administration

in having the recorded statement rated for its broadcast quality. After reading the contextual

scenario, participants were randomly assigned to one of the eight possible conditions.

were told that UK’s President is seeking recommendations about policy changes to be

implemented the following year (Spring, 2023), meaning that they, personally would be directly

affected by the policy. Participants assigned to the low personal involvement condition were told

the policy. Consequently, following prior research (Petty & Cacioppo, 1984), issue involvement

was manipulated solely by a change in the advocated date for implementation. After reading their

how important this issue was to them.

graduation. The message contained either strong or weak arguments and was delivered either in a

GAE or VE accent.

After listening to the recording, participants were first asked to respond to measures

assessing message persuasiveness. Following this, they were asked to evaluate the message

content (i.e., strength/quality) and the speaker (i.e.., status, solidarity). After completing these

items, they were then asked to answer standard demographic questions. Lastly, participants were

debriefed and told that the issue presented in the opening statement––the implementation of the

comprehensive exams–– was not actually being implemented next year or in the foreseeable

future at the University.

Participants

Volunteer student participants (n = 655) were recruited from the University of Kentucky

(4.3%), Native American (1.2%), and Other (1.4). Students were freshmen (59.7%), sophomores

size of 351 participants was required (Calculated using G*Power 3.1.9, a free-to-use software

used to calculate statistical power). After excluding participants who failed to meet inclusion

criteria (see above), the final sample (N = 347) was slightly under the sample size necessary for

sufficient power to detect small-to-medium effects.

Stimuli

Involvement Manipulation. Before message exposure, each participant was provided

with one of two scenario statements. Participants assigned to the high involvement condition

were told that UK’s recommendations about policy changes would be instigated the following

year (Spring, 2023), meaning that they themselves would have to take the exam and pass it in

involvement), meaning that they themselves would not have to take the exam or pass it in order

to graduate. Similar manipulations have been used extensively in past research (e.g., Petty, &

students (n = 130), participants in the high involvement condition (M = 5.31) reported the issue

as more personally relevant than participants in the low involvement condition (M = 3.71),

midpoint (i.e., 4), whereas the low involvement condition was marginally lower than the scale

midpoint, t(64) = -1.35, p < .091, d= -.17.

strong) in audio format. Each of the messages utilized an appeal advocating that seniors be

required to pass a comprehensive exam in their declared major before graduation. Overall, the

strong version of the appeal provided evidence (e.g., relevant data) supporting the argument. In

contrast, the weak version relied more on quotations and opinions than statistics and data to

support its argument.

The chosen stimuli messages were picked from a pool of statements acquired from past

literature (see Petty & Cacioppo, 1986). The arguments were pretested out of 5 possible

contextual messages comprising both strong and weak versions (total 10 messages). In order to

verify their relative strength, weakness, and comprehensibility, two pretests were conducted.

An initial pretest on a sample of undergraduate students (n = 258) sought to identify the

Message 5’s mean scores for the strong messages (M = 5.11) were significantly different from the

weak messages (M = 3.55), t(222)= -8.318, p < .001, d= -1.54. Overall, based on the largest

A second pretest built on the initial pretest’s findings by updating its issue involvement

statement. Due to this change, a second pretest on a sample of undergraduate students (n = 130)

sought to re-confirm that the choice of Message 5 was still a robust argument stimulus.

Subsequently, only two contextual message contexts were used for the second pretest (i.e.,

Messages 1 and 5; Appendix B). Pretest 2’s findings demonstrate that Message 1’s mean scores

for the strong message (M = 4.57) were significantly different from the weak message (M =

3.32), t(127)= -5.033, p < .001, d= -0.89. Notably, Message 5’s mean scores for the strong

message version (M = 4.82) were still significantly different from its weak message version (M =

2.92), t(127)= -7.508, p < .001, d= -1.32, displaying larger mean differences and stronger effect

matched-guise technique (MGT: Lambert, Hodgson, Gardner, & Fillenbaum, 1960). This method

involves using recordings (known as guises) from a single bidialectal speaker to produce all the

accented recordings. Variations on the MGT sometimes utilize multiple speakers to create the

stimuli creation, it suffers from interspeaker variation (e.g., pitch, rhythm). This variation

between recorded speakers has the potential to impact the targeted vocal manipulation (i.e., their

accent) under investigation. Subsequently, this study follows the matched-guise technique by

utilizing only a single speaker to create the accented recordings (i.e., guises) as this ensures that

each recording reflects only a difference in the manipulated accent.

All message recordings were produced by a male speaker. He was in his 30s and a native

Vietnamese and US English speaker. Participants were randomly assigned to listen to one of four

possible accented audio stimuli produced by the bidialectal speaker. The four recordings included

two recordings in a native, General American English (GAE) guise (i.e., strong and weak

message appeal) and two recordings in a foreign, Vietnamese English (VE) guise (i.e., strong and

weak message appeal). All recordings were of similar length (approx. 30s) and normalized at

70dB in post-production editing.

similarity, t(102) = 7.503, p = <.001, d = 1.47, and strength, t(102) = -12.604, p <.001, d = -2.47,

familiar (MVE = 3.47; MGAE = 4.69), less similar (MVE = 1.11; MGAE = 3.49), and more accented

was rated higher than the GAE guise in solidarity ratings (MVE = 5.31; MGAE = 4.60), t(102) = -

3.02, p = .003, d = .-0.59.

In sum, Pretest 3 confirmed that the chosen speaker was able to authentically adopt both

accented guises and that they were manipulated successfully. Findings were largely consistent

with expectations and prior literature (e.g., Dragojevic & Goatley-Soan, 2020; Fuertes et al.,

2012), as the VE guise was downgraded on most measures relative to the GAE guise. While

solidarity ratings were inconsistent with expectations, the relative increase in solidarity traits of a

given accent is neither unique nor problematic to the success of the manipulation. Sometimes,

end (see Appendix E for all proposed scales). Accordingly, this section introduces the dependent

measures assessing the central constructs of the experiment: source accent evaluation (i.e.,

status, solidarity, fluency) and message persuasiveness (i.e., attitude toward the issue, message

agreement); and lastly provide manipulation checks (i.e., message quality and issue

level of agreement with the message (i.e., to institute a comprehensive exam requirement for

seniors).

Attitude Toward the Issue. Evaluations toward comprehensive exams were measured

using a measure of general attitudes, which has yielded reliable results (Cronbach’s α = .97) in

past research (e.g., Ivanov et al., 2017). The measure includes seven semantic differential items

on a 7-point scale: bad/good, negative/positive, unfavorable/favorable, wrong/right,

unacceptable/acceptable, undesirable/desirable, and like/dislike. The scale yielded reliable

results across pretest 1 (α = .95) pretest 2 (α = .94), and the main experiment (α = .94).

indicate the extent to which they agree with the proposal using a 7-point scale (1= do not agree, 7

Source Accent Evaluation. After completing the message persuasiveness measures,

perceptions of the source’s accent were assessed using traditional and adapted measures

frequently used in language attitudes literature (see Dragojevic, Giles, Beck, & Tatum, 2017;

Zahn & Hopper, 1985). Accordingly, participants rated the speaker’s status and solidarity by

responding (1 = not at all; 7 = very) to five status (i.e., intelligent, educated, competent, smart,

and successful) and five solidarity (i.e., friendly, nice, pleasant, sociable, and warm) trait items.

source (M = 5.04, SD = 1.14) were attributed equal status.

H3b predicted that the VE-accented source will be attributed less solidarity than the

1.44) and GAE-accented source (M = 4.50, SD = 1.49) were attributed equal status. All

interactions involving source accent were nonsignificant.

However, there was a significant main effect of argument quality on solidarity

weak arguments were attributed more solidarity (M = 4.73, SD = 1.38) than sources who

delivered messages containing strong quality arguments (M = 4.30, SD = 1.48). The effects for

involvement were all nonsignificant (p = .411).

exams were more favorable: 1) under low involvement (M = 3.20, SD = 1.40) than high

involvement conditions (M = 2.86, SD = 1.43), 2) when the message contained strong (M = 3.36,

SD = 1.42) rather than weak arguments (M = 2.73, SD = 1.36), and 3) when the source spoke

with a GAE accent (M = 3.27, SD = 1.43) rather than VE accent (M = 2.80, SD = 1.39).

For agreement with the policy, an identical pattern emerged, with significant main effects

3.01, SD = 1.55).

condition (M = 4.24, SD = 1.25). Additionally, there was a significant main effect of source

elaborate when the message was delivered by the GAE-accented source (M = 4.67, SD = 1.23)

than a VE-accented source (M = 4.14, SD = 1.28). All effects involving message quality were

nonsignificant (p = .178).

Ability

Only a significant main effect of source accent on ability to elaborate emerged, F(1, 339)

when it was delivered by a VE-accented source (M = 3.97, SD = 1.51) than a GAE-accented

source (M = 4.52, SD = 1.35).

Depth

CHAPTER 5. DISCUSSION

This dissertation aimed to assess the impact of source accent and message quality under

varying listener involvement conditions. Under a low or high involvement contextual scenario,

student participants—all of whom were American nationals—listened to a recorded policy

American English (GAE) or Vietnamese English (VE) accented source. Findings provide clear

evidence that Americans reliably distinguished between native (i.e., GAE) and non-Anglo

he provided strong rather than weak quality arguments; and when the audience had low rather

than high involvement. Each of the key outcomes of this study—i.e., social categorization,

processing fluency, language attitudes, and persuasion—are discussed in turn.

Social Categorization

Within the US, native-accented speakers tend to be categorized as “American,” whereas

Consistent with H1, the majority of participants categorized the GAE-accented source as

‘American’ (63.8%) and the VE-accented source as ‘foreign’ (93.1%). This highly attuned and

with past literature (e.g., Kinzler, Shutts, & Correll, 2010) and the overall claim that accents are a

potent cue to social categorization (Lippi-Green, 2012; Rakić, Steffens, & Mummendey, 2011).

Moreover, this finding follows prior research by demonstrating that US participants readily

associate the GAE source with a native (American) identity and the VE source with a foreign

(non-American) identity (Dragojevic & Goatley-Soan, 2020). Consequently, the ideological

association between source accent and presumed nationality, along with the ability to accurately

differentiate between accents, provides the most plausible explanation for why US participants

were more likely to categorize the GAE source as American and the VE source as foreign or not

American. Stated differently, VE is a non-Anglo foreign accent, and GAE is a native US accent.

Different accents are indexical of various social identities; notably, a speaker’s accent is a

particularly salient marker to their presumed nationality (Shuck, 2004). US participants,

therefore, tended to ideologically equate VE with a foreign identity and GAE with a native

identity. Consequently, source accent functioned as a salient cue to social categorization, leading

participants in this study to be more likely to categorize the VE accented source as foreign and

the GAE source as American.

Processing Fluency

Previous research has shown that the more a person’s accent differs from one’s own, the

more difficulty one experiences processing that person’s speech (Cristia et al., 2012; Munro &

predicted that US participants would report lower processing fluency when listening to a VE-

than a GAE-accented source. In the present study, participants rated the VE accent as stronger,

H2, participants experienced more difficulty processing the VE-accented source’s speech than

the GAE-accented source’s speech. This finding is consistent with past research showing that

foreign-accented speech tends to disrupt listeners’ processing fluency relative to native-accented

speech (Cristia et al., 2012).

Additionally, although not predicted, source accent and message quality interacted to

influence listeners’ processing fluency. Although listeners experienced more difficulty

understanding VE than GAE speech, regardless of argument quality, the VE-accented source was

rated harder to understand when he delivered weak rather than strong arguments. In contrast,

processing fluency for the GAE-accented source was not influenced by argument quality. This

finding is important because it suggests that listeners’ experience of processing fluency is

influenced not only by a speaker’s accent, but also by message content. Past research has shown

that semantically coherent information is easier to process than incoherent information, and

messages that use concrete examples can also ease the cognitive processing of information

(Shulman & Sweitzer, 2020; Topolinski & Strack, 2009). Moreover, the processing fluency of

accented speech is known to be impacted by external (e.g., level of background noise) and

Language Attitudes

Past research has shown that Americans typically evaluate native US-accented English

speakers more favorably than non-Anglo foreign-accented English speakers on status and

solidarity traits (Dragojevic & Goatley-Soan, 2020; Fuertes et al., 2012). Two distinct processes

and VE accents were evaluated similarly on status and solidarity traits. These inconsistent

findings are especially surprising given that the VE-accented source was more likely to be

One possibility for these null findings is that US listeners’ stereotypes toward foreigners

has repeatedly demonstrated that Americans’ stereotypes toward non-Anglo foreigners are often

negative (Fuertes et al., 2012; Gluszek & Dovidio, 2010). Moreover, these negative attitudes are

especially evident towards speakers categorized as being from stigmatized countries/regions (i.e.,

Vietnam/Southeast Asia; Dragojevic & Goatley-Soan, 2020). A second explanation is that the

disruption in fluency was not significant enough, or was attributed to a cause other than the

speaker. These explanations are possible as a person’s processing fluency experience only

influences judgments when its impact is considered consequential to the primary judgment at

hand (Alter & Oppenheimer, 2009). Since people usually assign an occurrence to a single cause

rather than to multiple, it is possible that participants disregarded the disruptive impact of VE

speech or attributed its effects to an external cause (Dragojevic, 2019; Kelley, 1973;

Oppenheimer, 2006). A third possibility is that participants inferred who the speaker was based

on the study’s context (e.g., an educator or an administrator). Past literature has shown that

contextual markers can influence speaker evaluations by acting as the salient cue for social

understand than speech produced in native accents—which are more similar to one’s own.

disrupts listeners’ processing fluency may also depend on message content. Lastly, despite

finding differences in social categorization and fluency, no differences in language attitude

ratings (i.e., status, solidarity) were shown to emerge between the two accented sources.

Persuasion

The primary objective of this study was to examine if the effects of source accent and

message quality on persuasion are contingent on a receiver’s level of involvement. According to

the ELM, under low involvement conditions, people primarily use evaluations of source

characteristics (i.e., accent) to guide their assessment of a persuasive message, regardless of

message argument quality. Therefore, H4 predicted that, under low issue involvement, a

VEaccented source would be less persuasive than a GAE-accented source, regardless of the

quality of their message arguments. Under high involvement conditions, however, the ELM

suggests that a person’s assessment of a persuasive message should predominantly depend on the

quality of its arguments, not on their evaluations of source characteristics (Petty & Cacioppo,

1986b).

Notwithstanding this typical understanding, the ELM’s multiple role hypothesis also indicates

that source characteristics can, in some contexts, not only function as peripheral cues but in

moderator. Consequently, H5 predicted that under high issue involvement conditions, strong

arguments would be more persuasive than weak arguments, but only when the source spoke with

a GAE accent. Based on this, a three-way interaction between source accent, message content,

persuaded when the source spoke with a GAE than a VE accent; when the message contained

strong rather than weak arguments; and when they had low rather than high involvement on the

issue. Each of these main effects is discussed in turn.

Effects of Source Accent

Policy messages delivered by the GAE source were more persuasive than policy

messages delivered by the VE source, regardless of argument quality and listener involvement.

This finding is consistent with previous research demonstrating that persuasive appeals presented

by native US accented speakers tend to be more convincing than those delivered by non-Anglo

foreign-accented speakers (DeSheilds, Kara, & Kaynak, 1996). The primary explanation for this

hinder persuasion (Dragojevic et al., 2018; Morales, Scott, Yorkston, 2012). However, in the

present study, the GAE and VE sources were rated equally on status and solidarity traits. Thus,

Recent research has shown that various forms of metacognition can impact persuasive

outcomes (Brinol & Petty, 2009). Notably, a person’s processing fluency can act as a

In brief, the unimodel, based on Kruglanski’s (1989) lay epistemic theory, suggests that

In contrast to the ELM’s dual-process view––that message arguments and heuristic/peripheral

cues are functionally distinct with a receiver’s level of elaboration determining which to privilege

for message evaluation––the unimodel proposes that they are functionally equivalent during

persuasive message processing (Kruglanski et al., 2014). In this way, the unimodel proposes that

both message arguments and peripheral cues make up what is referred to as evidence—

information relevant to a judgment (Kruglanski et al., 2006). The perceived

“argument strength” of information considered evidence is based on the degree of a receiver’s

belief that the particular information is relevant to the judgment (Kruglanski et al., 2006). In

other words, the more firmly a receiver believes that they can draw conclusions from the given

piece of information, the more convincing it will be as evidence.

Guided by this understanding, under high elaboration conditions, source accent and

message quality (i.e., given information) both likely served as information relevant to

Understanding how source accent impacts persuasion is not only important theoretically,

foreign accent rather than a native accent tend to be less persuasive, even when participants do

not readily attribute negative stereotype-based judgments to them. It appears that this continued

persuasive downgrading occurs because non-Anglo foreign-accented speech reduces listeners’

processing fluency and lowers the perceived strength of their message arguments. Together, these

findings extend the understanding of why foreign-accented speakers are routinely perceived as

less persuasive than their native-speaking counterparts, even in high elaboration contexts: Their

persuasive messages are not primarily evaluated on their informational merits but rather are

evaluated in conjunction with perceptions of their accented speech. Consequently, this empirical

investigation adds further support to the already wide range of social and professional settings

where foreign-accented speakers encounter prejudice and discrimination in the US (Lippi-Green,

2012). This study’s findings not only help reveal negative societal processes at work, but also

have practical implications for companies or organizations using foreign-accented experts or

speakers to deliver persuasive messages.

credible evidence in a persuasive message should help mitigate the difficulty receivers may

experience when processing a spoken message delivered in foreign-accented speech. This

increase in receivers’ processing fluency may also help to reduce negative feelings and improve

While this dissertation certainly constitutes a valuable step forward in understanding the

impact of source accent on persuasion, there are still areas for future research to investigate and

limitations that provide room for improvement. First, future researchers should investigate the

effects of source accent and message quality with participants under explicitly high and low

levels of elaboration. As previously noted, participants in this study were likely experiencing

more extensive levels of elaboration regardless of their level of involvement. Therefore, it would

be beneficial to see in a single experiment whether source accent and message quality interplay

at these lower levels as predicted by the ELM. Second, future research should explore whether

these findings generalize to other persuasive contexts and variables that interact with a source’s

accent. For example, the relative persuasiveness of accented speakers has been shown to be

impacted by certain message topics, advocated positions, and receiver group identities

(DeShields & Kara, 2011; DeShields et al., 1997; Giles et al., 1995). Lastly, findings from the

supplemental analyses indicate that source accent may also impact persuasion via a participant’s

motivation and ability to process messages, indicating that it may function as an elaboration

a single non-Anglo foreign-accented variety. Using a single foreign accent raises concerns about

how findings generalize to other foreign accents. For example, not all foreign-accented speakers

are evaluated equally, as an evaluative hierarchy tends to emerge (Dragojevic & Goatley-Soan,

non-Anglosphere, Southeast Asia) are typically evaluated less favorably than speakers associated

with non-stigmatized groups (i.e., Anglosphere, Western Europe).

foreign accents. Nonetheless, future research should attempt to replicate these results using a more

increase or decrease student participant involvement. While it is not clear why this study’s

methodological replication of issue involvement did not obtain the desired low elaboration effect,

it is possible that an alternative involvement manipulation may have more success. For example,

other experiments have manipulated involvement by changing the school at which the exams will

be implemented: theirs or another (e.g., Petty & Cacioppo, 1979). Accordingly, to fully

understand source accent effects under both processing routes in the same experiment, future

research could attempt to employ more successful and/or varying means for producing lower

Fourth, this study predicted a three-way interaction because source accent was theorized

previously mentioned, the foreign-accented source significantly reduced receivers’ ability to

process messages relative to the native-accented source. Even so, ability in the foreign-accented

source condition remained relatively high and not significantly lower than the scale midpoint

(i.e., 4), t(172) = - .27, p = .395, d = - .02. ELM literature suggests that for a specific variable to

redirect processing to the alternate route, it must considerably impact a receiver’s ability to

process information (i.e., medium to high levels, see Petty, Wells, & Brock, 1976). Accordingly,

the foreign-accented source may not have reduced ability substantially enough to prevent central

processing of the message. Based on this understanding, had the foreign-accented source spoken

with an even stronger accent, perhaps it would have disrupted ability even further and initiated

peripheral processing; future studies should examine this possibility.

It is also possible that the foreign-accented source used in the present study would have

had a more substantial impact on receivers’ ability to process the message had the participant

sample been different. Past research has shown that prior exposure to a particular accent can

those who have had minimal exposure to foreign-accented speakers, perhaps the foreign-

accented speaker would have caused a more substantial disruption to their ability to process the

utilize a more diverse sample of participants.