Is Your Accent Right for the Job? A Meta-Analysis on Accent Bias in Hiring Decisions

Abstract

Standard-accented job candidates are perceived as more hireable than non-standard-accented candidates. Two broad perspectives have emerged as to what drives this effect: (a) that it is a pragmatic response to the perception that non-standard accents can impede job-relevant communication (processing fluency explanation) and/or (b) that non-standard accents signal “otherness” and candidates are devalued as a result (prejudice explanation). This meta-analytic integration of 139 effect sizes (N = 4,576) examined these two perspectives. Standard-accented candidates were considered more hireable than non-standard-accented candidates (d = 0.47)—a bias that was stronger for high communication jobs. Other findings, however, are difficult to explain from a processing fluency explanation: candidates’ relative comprehensibility was not a significant moderator of hiring bias. Moreover, the degree of accent bias was associated with perceptions of the candidates’ social status, and accent bias was particularly pronounced among female candidates and for candidates who spoke in foreign (as compared with regional) accents.

Keywords

aversive prejudice discrimination language attitudes processing fluency job recruitment

In 2019, 272 million people were living in a country other than their place of birth (United Nations Department of Economic and Social Affairs, 2019). Among the top motivations for migration is seeking better job opportunities (Yakushko et al., 2008). Yet, entering the workforce is not an easy process for international migrants (Ahmad, 2020). One challenge is differences in spoken accent; that is, one’s manner of pronunciation as a result of grammatical, syntactical, morphological, and lexical aspects of one’s native language (Giles, 1970). Indeed, when two equally qualified job candidates differ in accent, the standard-accented candidate (i.e., the candidate who speaks with the accent that is generally accepted and institutionalized as the way of speaking a given language) is more likely to be hired than the non-standard-accented candidate (e.g., Deprez-Sims & Morris, 2013).

What psychological processes drive this accent-based hiring bias has been a topic of significant debate. Two broad arguments have emerged: First is a processing fluency explanation—that it is a pragmatic response to the fact that non-standard accents can impede communication, and by extension can make employees less effective in their job. Processing fluency refers to the relative ease or difficulty with which information is processed (Dragojevic & Giles, 2016). The processing fluency perspective on accent-based biases reflects the idea that non-standard accents are cognitively effortful to process—that is, they are more difficult and require more cognitive resources to understand relative to standard accents (Floccia et al., 2009; Munro & Derwing, 1995). The poorer processing fluency associated with non-standard accents, in turn, leads to negative consequences for non-standard speakers (Hideg et al., 2022).

Second, and in contrast, is a prejudice explanation—that non-standard accents signal “otherness” and candidates are devalued as a result. The prejudice perspective on accent-based biases is rooted in social categorization and stereotypes (Ryan, 1983). From accent, listeners can infer the speaker’s social group membership(s) (e.g., their gender, ethnicity and/or nationality; Ko et al., 2006; Rakić et al., 2011); this, in turn, activates stereotypes associated with those social groups which are then attributed to the speaker (Giles & Watson, 2013). For non-standard-accented speakers, the activation of (often) negative stereotypes has commonly been used to explain their downgrading relative to standard speakers (Gluszek & Dovidio, 2010; Hideg et al., 2022).

Disentangling these two explanations is not easy because of the principle of aversive prejudice (Dovidio & Gaertner, 1986; Hodson et al., 2002). From this perspective, people are motivated to maintain a self-image as egalitarian and so will only discriminate across race, sex, and so forth when they can rationalize their decisions as being non-prejudiced. In the case of accent bias, it could be that people cognitively rally around a processing fluency explanation for why they should not hire a candidate with a non-standard accent, when in fact they may be driven (perhaps unconsciously) by broader implicit prejudices. Indeed, while prior research has demonstrated that some adults may constrain their explicit accent-based biases and even display “over-correction” tendencies by favoring non-standard speakers (Pantos & Perkins, 2013; Roessel et al., 2020), their implicit prejudices reflected in their automatic and immediate reactions (Wittenbrink & Schwarz, 2007) reveal favoritism for standard speakers (e.g., Pantos & Perkins, 2013; Romero-Rivas et al., 2021).

In trying to uncover the psychological drivers of accent bias in hiring decisions, one line of research has manipulated the communication demands of the job that the candidates are applying for (e.g., Hosoda & Stone-Romero, 2010; Timming, 2017). For example, these studies have compared candidate hireability in customer-facing roles (e.g., customer service representative) versus non-customer-facing roles (e.g., data entry clerk). Another line of research has examined how hiring decisions are related to the candidates’ perceived qualities, such as their comprehensibility, accent strength, intelligence, and friendliness (e.g., Carlson & McHenry, 2006; Deprez-Sims & Morris, 2010, 2013; Goatley-Soan & Baldwin, 2018; Hansen et al., 2014, 2018; Roessel et al., 2019, 2020).

Although these findings offer important insights, individual studies may struggle to provide comprehensive answers to how differences in perceived qualities actually relate to hiring decisions and, ultimately, the question of what drives accent biases in hiring. This is partly because of the porous boundaries between the two explanations described earlier; that is, if accent bias is greater in jobs with high communication demands, it is difficult to detect whether this is a reflection of processing fluency concerns or whether it is a reflection of aversive prejudice (or both). Further complicating the picture is the fact that individual studies typically operate in individual settings—that is, they often examine only a small selection of accents with participants from a single geographical region. In this way, perceptions of the candidates may be heavily influenced by any stereotypes or idiosyncratic factors attached to individual candidates (or groups of candidates) from particular regions used in single studies.

To help overcome these problems, we conducted a meta-analysis of the studies on accent bias in hiring. Doing so has the usual benefits of offering a bird’s-eye view of the field across multiple individual candidates, accents, countries, and types of employment. Previous research has shown a mix of findings: some show large accent effects, some show weak effects, and some show no effects. Meta-analyses help identify moderators that can explain this between-study variability. Importantly, doing so may also shed light on the elusive question of whether accent biases in hiring are driven by processing fluency versus prejudicial concerns.

Based on previous literature, we expected an accent bias overall such that standard-accented candidates are considered more hireable than non-standard accented candidates. To examine how much accent biases in hiring are driven by processing fluency versus prejudicial concerns, we examined several moderating variables. We discuss these moderators below.

Moderators Related to Processing Fluency

Accentedness

A single “hello” is all it takes for a listener to register that a speaker has a standard or non-standard accent (Baugh, 2000). Not only are we proficient at detecting when someone speaks with a non-standard accent, but we can also readily identify their “accentedness”—that is, the relative strength of someone’s non-standard accent relative to the standard accent (e.g., Roessel et al., 2019).

Furthermore, accentedness directly influences judgments on how easy or difficult one is to understand (i.e., processing fluency). The impact of accentedness on processing fluency has, in turn, been linked to trait evaluations of the speakers (Dragojevic et al., 2017; Lev-Ari & Keysar, 2010). For example, Dragojevic and colleagues’ (2017) demonstrated that greater levels of accentedness led to higher reported disruptions in listeners’ processing fluency. This, in turn, was associated with more negative evaluations of personal characteristics of speakers with strong foreign accents relative to mild foreign accents. These links between accentedness, processing fluency, and trait evaluations are in line with the processing fluency perspective: Standard-accented candidates may be evaluated more positively than non-standard-accented candidates due to differences in ease of understanding based on accentedness.

A large proportion of studies on accent bias in hiring decisions have asked participants to rate each candidates’ level of accentedness. Not surprisingly, non-standard-accented candidates are typically rated as more strongly accented compared with standard-accented candidates (e.g., Deprez-Sims & Morris, 2013; Roessel et al., 2019). In the majority of studies, however, the predominant motivation for including the accentedness measure has been to ensure that participants registered the non-standard accent (i.e., a manipulation check). Few studies have investigated accentedness as a predictor of hiring biases. Accordingly, in the present meta-analysis, we tested for overall differences in perceived accentedness between standard and non-standard-accented candidates and then examined the extent to which these perceived differences in accentedness impacted accent-based hiring decisions.

Comprehensibility

In language attitudes research, the term “comprehensibility” is used to discuss two distinct types of processing fluency: objective comprehensibility (commonly known as “intelligibility”) which refers to how much you can actually understand someone and subjective comprehensibility which refers to how easy it feels to understand someone. In this meta-analysis, we use the term “comprehensibility” to refer to subjective comprehensibility given that this is how comprehensibility was operationalized and measured in the included studies. Furthermore, studies that have compared the two have found that subjective comprehensibility (more so than objective comprehensibility) impacts speaker evaluations (e.g., Hansen & Dovidio, 2016).

Non-standard accents are typically perceived to be less comprehensible than standard accents (Munro & Derwing, 1995, 1999). This perceived difficulty in processing non-standard accents is often viewed as a major barrier to communication and has profound ramifications on non-standard-accented speakers with regard to legal systems, access to housing, and employment (Gluszek & Dovidio, 2010).

To investigate the impact of job candidates’ perceived comprehensibility on hiring, numerous studies have asked participants to judge the relative ease of understanding standard and non-standard-accented candidates. If processing fluency is at the crux of accent bias in hiring decisions, we should find that the relative comprehensibility of the standard accent over the non-standard accent moderates the strength of the bias. The extent to which perceptions of comprehensibility account for accent-based hiring biases, however, is not clear cut: Some studies have found large differences between standard and non-standard candidates in perceived comprehensibility (e.g., Hosoda & Stone-Romero, 2010; Roessel et al., 2020) and that these perceptions are associated with hiring decisions (Deprez-Sims & Morris, 2013). In contrast, other studies have provided evidence that comprehensibility perceptions do not account for hiring biases (Deprez-Sims & Morris, 2010).

One explanation for these mixed findings may stem from the differences in types of non-standard accents used across studies. Indeed, non-standard accents vary in relative comprehensibility (Munro & Derwing, 1999). Furthermore, given that most individual studies examine only a single or very small selection of non-standard accents, it is difficult to determine within an individual study the degree to which comprehensibility perceptions relate to hiring biases more broadly. By aggregating findings from studies in the literature that have used a range of accent varieties, our meta-analysis aimed to determine the extent to which comprehensibility impacts hiring decisions between standard and non-standard-accented candidates.

Job Communication Demands

Another potential moderator that speaks to the processing fluency perspective is the level of communication required for the job. If communication barriers explain why non-standard-accented candidates are less preferred over standard-accented candidates, we would expect that this bias against non-standard-accented candidates would be reduced when candidates are applying for jobs that require less communication.

Prior research has compared accent-based hiring biases in jobs that require high levels of communication (e.g., customer service representative) versus jobs that require low levels of communication (e.g., data entry clerk). This research suggests that non-standard-accented candidates are perceived to be less hireable in roles with high communication demands relative to roles with low communication demands (Hosoda & Stone-Romero, 2010; Timming, 2017). Nevertheless, preference for standard-accented over non-standard-accented candidates was still evident in jobs with low communication demands (Timming, 2017).

To date, only two studies have directly compared hiring decisions for both standard and non-standard-accented candidates when applying for jobs with both high and low communication demands (Hosoda & Stone-Romero, 2010; Timming, 2017). While other studies have not specifically manipulated communication demands, many have identified the level of communication required in a single job context (Nguyen, 2015; van de Wouw, 2020). Therefore, our meta-analysis was able to synthesize across studies and job types to determine the effect of job communication demands on hiring success.

Moderators Related to Prejudice

Status and Solidarity

Non-standard accents have been associated with a range of negative stereotypes that underpin prejudice (Gluszek & Dovidio, 2010). For example, non-standard-accented speakers are judged as less believable (Lev-Ari & Keysar, 2010), less loyal (Edwards, 1982), and less intelligent (Bradac, 1990; Lindemann, 2003). Research on accent-based biases highlight that these stereotypes tend to converge on two dimensions: status and solidarity (also referred to as “competence” and ‘warmth’; Fiske et al., 2002; Gluszek & Dovidio, 2010). Perceptions of a speaker’s status include evaluations of the speaker’s competence, intelligence, ambition, and social class, while perceptions of a speaker’s solidarity include evaluations of the speaker’s friendliness, warmth, likeability, and trustworthiness. One’s accent is such a powerful communicator of traits related to these dimensions that children as young as nine associate standard accents with higher status traits (e.g., “smart” and “in charge”) relative to non-standard accents (Kinzler & Dejesus, 2013).

To better understand how stereotypes relate to accent-based hiring biases, studies have captured participants’ evaluations of candidates along dimensions of status and solidarity. In general, these studies suggest that non-standard-accented candidates are perceived to be lower in status than standard-accented candidates (Gluszek & Dovidio, 2010). On solidarity traits, however, findings are less clear: While non-standard-accented candidates have been found to be downgraded on this dimension (Fuertes et al., 2012), sometimes non-standard-accented candidates are rated equally, if not higher, in solidarity than standard-accented candidates (e.g., Hosoda et al., 2012; Roessel et al., 2020). The variability in findings on solidarity traits may depend on competing processes that have been documented to impact solidarity judgments in particular—that is, due to compensation and in-group versus out-group identification mechanisms (Roessel et al., 2019). For example, non-standard accent varieties may evoke higher solidarity ratings as compensation for lower status ratings (Yzerbyt et al., 2005) and, more specifically, regional accent varieties of one’s own native language may evoke higher solidarity ratings due to positive in-group identification (Ryan, 1983).

Our meta-analysis capitalized on the fact that many studies have examined participants’ perceptions of standard- and non-standard-accented candidates along these dimensions and did so consistently on Likert-type scale ratings. Accordingly, we aggregated findings across previous studies to examine links between the degree of accent-related bias on dimensions of status and solidarity and the extent of hiring biases. Based on the prejudice perspective—which suggests that accent-based bias is rooted in social categorization and stereotypes—we expected standard-accented candidates to be rated higher in status overall but perhaps equal in solidarity relative to non-standard-accented candidates. Furthermore, we expected that the greater perceived differences between candidates on status and solidarity ratings would moderate hiring decisions such that the more the standard-accented candidate was considered higher in status and solidarity relative to the non-standard-accented candidate, the more the standard-accented candidate would be hired.

Sociodemographic Characteristics

The way one speaks communicates a great deal of information about that speaker. For example, from a brief audio clip, listeners can determine the speaker’s gender (Ko et al., 2006), make inferences about where they are from (Derwing & Munro, 2009), and infer their ethnicity (Rakić et al., 2011). Consistent with the prejudice perspective, therefore, it is possible that other biases could manifest in hiring decisions that are the result of group-based social categorizations and stereotypes triggered by the candidate’s accent.

Our meta-analysis examined whether hiring bias was greater when the candidate possessed sociodemographic characteristics that made them more vulnerable to prejudice; specifically, their gender and their non-nativeness (i.e., whether candidates spoke with regional versus foreign accents). In this meta-analysis, we considered “regional” accents to include native variations of a language due to dialects (e.g., Northern American-accented English versus Southern American-accented English) or ethnic differences (e.g., standard American-accented English versus African American English varieties) within a country as well as native variations of a language between countries (e.g., American-accented English versus British-accented English). We considered “foreign” accents to include variations of a language as a result of the candidate having a different native language (e.g., a non-standard-accented candidate who speaks English with a Chinese accent due to their native language being Chinese).

In many settings—including the workplace—cues to ethnicity and gender can be met with discrimination (Barak, 2022; Nkomo & Kinahan, 2015). Strong norms against race- and gender-based discrimination, however, are prevalent in society. In contrast, minimal norms exist against accent-based discrimination (Gluszek & Dovidio, 2010). Consequently, it could be predicted from an aversive prejudice perspective that broader racial-ethnic and gender prejudices may emerge in hiring decisions given that one’s accent (and associated processing fluency concerns) may provide a socially acceptable excuse to not hire a gender and racial-ethnic minority candidate. Specifically, we expected the accent bias in hiring might be more pronounced when the candidates’ non-standard accent signaled non-nativeness (as compared with native differences) and when the candidate was female (compared with male).

Present Meta-Analysis

In sum, this meta-analysis aimed to shed light on the elusive question of whether accent biases in hiring are driven by processing fluency versus prejudicial concerns. To do so, we provided the first statistical synthesis of the literature on accent-based hiring biases. Furthermore, we investigated moderators that relate to concerns around processing fluency (e.g., candidates’ accentedness, comprehensibility, and the communication demands of the jobs candidates are applying for) as well as moderators that could reveal prejudices within accent-based hiring decisions (e.g., evaluations of a candidate’s status and solidarity, candidate’s gender, and whether candidates spoke with a foreign or regional accent).

Method

Transparency and Openness

A protocol for this meta-analysis following the PRISMA(P) statement (Moher et al., 2015) was developed prior to commencing the literature search and screening process. All materials associated with the study, including the protocol, datafiles, and scripts for analysis, are publicly available on the Open Science Framework (https://osf.io/t4jrp/).

Literature Search

We conducted a literature search on studies that have investigated the effect of accent on hiring decisions. The search was completed in May 2021. To identify studies eligible for inclusion, we searched the following databases: Web of Science (Thomson Reuters), PsycINFO (American Psychological Association), Scopus (Sci-Verse), ABI Inform (ProQuest) and ProQuest Dissertations and Theses Global (ProQuest). We used the following two keyword clusters: first, “native,” “foreign,” “regional,” “local,” “standard,” “non-native,” “non-regional,” “non-local,” and “non-standard” were searched in combination with “accent,” “accentedness*,” “dialect*” and “language.” This search was combined with our second cluster of terms: “hiring,” “employment,” “job,” “applicant,” “candidate,” “decision*,” “recommendation,” “outcome,” “suitability,” “recruitment,” “employability,” “hirability,” “hireability” and “selection decision*.” In addition, we conducted backward citation searching and contacted authors of recent publications requesting unpublished data. To search for gray literature, we screened the first 20 pages (200 results) in Google Scholar (Haddaway et al., 2015) with an abbreviated search (see Supplementary Materials for search details). Figure 1 displays details of the sequence for the identification of eligible studies.

Figure 1.

Flow diagram of sequence of exclusion and inclusion of eligible studies based on the Preferred Reporting Items for Systematic and Meta-Analytic Reviews (PRISMA; Moher et al., 2009).

Inclusion Criteria

Eligible studies were selected based on the following inclusion criteria:

Studies were in the context of job recruitment whereby job candidates were evaluated for real or hypothetical employment.

Studies used a quantitative design.

Studies included a measure of candidate hireability (e.g., hiring decision, hiring recommendation, likelihood of hiring, and job suitability).

Studies included a hiring evaluation between at least one standard-accented candidate and at least one non-standard-accented candidate. Standard speakers were those who spoke with an accent generally accepted, taught in schools, and institutionalized as the way of speaking a given language (i.e., American-accented English in the United States or German-accented German in Germany). A small subset of the literature (7 effect sizes) contrasted candidates who spoke English with a native accent versus those who spoke English with a non-native accent but in the accent of the region in which the study was conducted (e.g., a study conducted in Germany with an American-accented candidate versus a German-accented candidate who both spoke in English). Due to the inherent difficulty of determining which speaker is “standard” in these cases, these effect sizes were excluded.

The defining difference between candidates was the accent they spoke—that is, studies primarily manipulated vocal characteristics pertaining to pronunciation-linked variables of the candidates. For 17 effect sizes, the defining difference was the candidate’s accent, but additional non-visual cues to ethnicity were provided (i.e., via the candidates’ names or country of origin). We included these 17 effect sizes in the meta-analysis, given that assumptions about candidate ethnicity could be made through one’s accent alone, even in the absence of specific ethnicity information. A moderator analysis confirmed that the mean effect sizes did not differ for studies with additional cues to ethnicity versus studies without additional cues to ethnicity, F(1,137) = 0.01, p = .91. Studies where the candidate’s accent was confounded by another cue (e.g., manipulations of candidate appearance, attractiveness, or employment qualifications) were not included. In addition, studies that used interventions aimed at reducing accent-based biases (e.g., informing participants to be wary of accent-related prejudices prior to the task) were not included.

Studies reported precise statistics which could be converted to effect sizes.

As displayed in Figure 1, 27 studies conducted between 1984 and 2021 met these criteria, contributing 139 individual effect sizes. The full references for the 27 studies included in this meta-analysis are marked by asterisks in the References section.

Coding

We investigated the influence of several moderating variables, including perceptions of the candidate’s: (a) accentedness; (b) comprehensibility; (c) status; and (d) solidarity. In addition, we coded for: (e) job communication demands (low versus high); (f) candidate gender (male versus female); and (g) linguistic cue (foreign accent versus regional accent). These moderating variables, along with the corresponding sample sizes and effect sizes, were coded by the first and third authors. Each study was also coded for the following secondary variables: (a) time exposed to candidates (mean seconds exposed); (b) hiring experience of participants (novice versus expert); (c) job status (low versus high); article characteristics such as (d) year of publication; and (e) publication status; as well as sample characteristics (f) sample size; (g) proportion of males; and (h) mean age. More detailed information on coding is provided in the Supplementary Materials.

Interrater reliability between the two coders was high for each of the categorical measures: linguistic cue (κ = .97), candidate gender (κ = .99), hiring experience (κ = 1.00), job status (κ = 1.00), job communication demands (κ = 1.00), and publication status (κ = 1.00). The intraclass correlations were excellent for continuous variables: accentedness (1.00), comprehensibility (.98), status (1.00), solidarity (.99), time exposed to candidate (1.00), sample size (1.00), proportion of males (1.00), mean age (1.00), year of publication (1.00), and effect size (.96). All disagreements were discussed to reach agreement between the two coders.

Sample Characteristics

A total of 4,576 participants (57.65% female, M_age = 32.15 years) took part in the eligible studies that contributed data to this meta-analysis. Samples were from Germany, Netherlands, New Zealand, Portugal, United Kingdom, and United States. Participants spoke the following languages: Dutch, English, French, German, Italian, Portuguese, and Spanish. The majority of studies that reported participant ethnicity (17 out of 27) involved a variety of ethnicities. Participants came from the following ethnic backgrounds: African American, Asian, Caucasian, Continental Africa, Dutch, European North American, German, Hispanic, Latino, Middle Eastern, Native American, Pacific Islander, Portuguese, Spanish, and mixed ethnicity.

Stimuli Characteristics and Experimental Design

Job candidates in the included studies spoke with a variety of standard English-speaking accents (e.g., American-accented, British-accented, New Zealand-accented, and South African-accented English), European-accented English (e.g., Dutch-accented, French-accented, German-accented, Russian-accented, Spanish-accented English), Asian-accented English (e.g., Chinese-accented, Indian-accented, and Japanese-accented English), and Mexican American-accented English. Included studies also featured standard German and other European-accented German (e.g., British-accented, Bulgarian-accented, Russian-accented, and Turkish-accented German). The majority of studies (19 of 27) used the “verbal guise” procedure, meaning different speakers were used to produce the standard- and non-standard-accented stimuli, while seven studies used the “matched guise” procedure, meaning the same bidialectal speaker produced both accent varieties. One study used both procedures in separate experiments (Rakić et al., 2011).¹ In the majority of studies (16 of 27), a between-subjects approach was taken—that is, participants either heard and evaluated a standard-accented candidate or a non-standard-accented candidate. Ten studies used a within-subjects approach, where participants heard and evaluated both a standard and non-standard-accented candidate. One study used both approaches across separate experiments (Roessel et al., 2020).² Specific details on stimuli characteristics for each study can be found on the Open Science Framework.

Statistical Analyses

We used standardized mean difference (Cohen’s d) to examine the effect of candidate accent on hiring decisions, whereby a positive Cohen’s d implied favorability to hire the standard-accented candidate over the non-standard-accented candidate. No studies provided Cohen’s d, so the means, standard deviations, and test statistics were used to calculate effect sizes and corresponding variances. In studies where means and standard deviations were provided, we followed Viechtbauer’s (2020) recommendations for calculation. In studies where only test statistics (e.g., t-tests, ANOVAs) were provided, we calculated effect sizes using the Comprehensive Meta-Analysis (CMA) software (Boreinstein et al., 2005).

We employed a three-level meta-analytic approach to calculate the overall effect of candidate accent on hiring decisions and investigate the influence of potential moderators on this effect. In contrast to the traditional meta-analytic approach which only accounts for sampling variance and between-study variance, the three-level meta-analytic approach accounts for variance in the dataset at three levels: sampling variance (Level 1), within-study variance (Level 2), and between-study variance (Level 3; Assink & Wibbelink, 2016). The three-level approach was most suitable for our dataset given that all but seven studies reported multiple, dependent effect sizes—that is, by taking the three-level approach, we were able to estimate the within-study variance on our overall effect and investigate the influence of potential moderating variables at this level (Cheung, 2015).

Analyses were conducted in the R environment (version 3.6.3; R Core Team, 2020) using the rma.mv function of the Metafor package (Viechtbauer, 2015). We employed a multilevel random effects model, used the restricted maximum likelihood estimation method to calculate all model parameters, and the t-distribution to test individual regression coefficients (Knapp & Hartung, 2003).

Sensitivity analyses

A leave-one-out analysis did not reveal any outlying studies. We also screened for extreme effect sizes with standardized scores larger than 3.29 or smaller than −3.29 (Tabachnik & Fidell, 2013). Two outliers were identified. The same overall pattern of findings emerged regardless of whether we included or excluded these outliers. Findings including all effect sizes are reported below.

Tests for publication bias

Methods to assess the presence and severity of publication bias have been developed for meta-analysis (Duval & Tweedie, 2000; Egger et al., 1997), but these methods have yet to be reliably adapted for three-level meta-analytic approaches. Therefore, to conduct publication bias tests on our dataset, we first aggregated the effect sizes from each published study included in our meta-analysis. To avoid false-negative findings, we only included data from published studies (i.e., we excluded unpublished studies from our dataset) in our tests for publication bias. We created a funnel plot of effect sizes plotted against its precision from each of the published studies. Asymmetry in the funnel plot, driven by few studies with small samples that yielded nonsignificant results, suggests publication bias. To statistically assess for funnel plot asymmetry, we used two methods: (a) following Egger et al.’s (1997) procedure, we regressed the standardized effect size on precision of the effect size, whereby a significant intercept suggests publication bias; and (b) we employed Duval and Tweedie’s (2000) trim-and-fill procedure that estimates the number of missing studies contributing to funnel plot asymmetry and provides an adjusted mean effect size by imputing values from the hypothetical missing studies.

As shown in Figure 2, the funnel plot appears to have minor asymmetry on the upper left and lower right quadrants where effect sizes appear to be missing. The statistical tests of funnel plot asymmetry, however, revealed publication bias was unlikely. Egger’s test for funnel plot asymmetry was nonsignificant, t(16) = 0.36, p = .73, and the trim-and-fill procedure estimated zero missing studies.

Figure 2.

Funnel plot plotting Cohen’s d against the standard error from the 18 published studies included in the meta-analysis.

Results and Discussion

Overall Effect of Candidate Accent on Hiring Decisions

The overall effect of candidate accent on hiring decisions was first calculated from the 139 available effect sizes. This analysis yielded a significant effect size, d = 0.47; 95% confidence interval (CI): [0.25, 0.69], t(138) = 4.15, p < .001, the magnitude of which would be considered rather large for social psychological studies (Lovakov & Agadullina, 2021). Overall, participants exhibited significant hiring bias, such that standard-accented speakers were favored over non-standard-accented speakers.

Heterogeneity analyses

One-tailed log-likelihood-ratio tests revealed that there was significant variance in effect sizes reported at the within-study level (p < .001) and the between-study level (p < .001). Based on Cheung’s (2014) formula, sampling variance, within-study variance, and between-study variance were estimated to account for 1.99%, 47.30%, and 50.71% of the total variance, respectively. Given the large proportion of variance estimated at the within- and between-study levels, moderator analyses were conducted to further investigate this variance.

Moderator analyses

Prior to analyses, we mean-centered all continuous moderating variables and coded all categorical moderating variables into dichotomous dummy variables. To investigate the significance of each moderator on the effect of candidate accent on hiring decision, we conducted a series of omnibus tests of a null hypothesis based on the F distribution. Table 1 provides an overview of the moderators. Table 2 summarizes the number of effect sizes included (which varied between analyses) and statistical output for each analysis.

Table 1.

Overview of Primary Moderators Including Definitions and Operationalizations.

Moderator	Definition and operationalization
Processing Fluency Moderators
Accentedness	Accentedness refers to the relative strength of the non-standard accent relative to the standard accent. In the included studies, accentedness was measured via 7-point Likert-type scales with responses ranging from no accent/standard accent to noticeable accent/strong accent, or agreement with statements such as, “The speaker has a strong foreign accent” ranging from completely disagree to completely agree.
Comprehensibility	Comprehensibility refers to how easy it feels to understand someone (i.e., subjective comprehensibility). In the included studies, comprehensibility was typically measured via 7-point Likert-type scales with responses ranging from very difficult to understand to easy to understand, or agreement with statements such as, “I found the speaker easy to understand” ranging from strongly disagree to strongly agree.
Job communication demands	This moderator examined whether the strength of accent bias in hiring differed according to whether the job had low or high communication demands. Effect sizes were only included in this moderator analysis if explicit information on the communication demands of the job that the candidates were being considered for was provided. The majority of the included studies pretested for perceived communication demands to confirm low versus high communication demands were manipulated successfully. Others provided explicit information on the level of communication required in a single job context.
Prejudice Moderators
Status	Status refers to perceptions of one’s social standing in society. In the included studies, perceived status of the candidate was measured using multiple traits, including competence, intelligence, social class, confidence, ambitiousness, competitiveness, independence, and wealth. These evaluations were typically measured via 5- to 7-point Likert-type scales (ranging from completely disagree to completely agree) or 7-point semantic differential scales (e.g., incompetent—competent, uneducated—educated).
Solidarity	Solidarity refers to perceptions of one’s good nature. In the included studies, perceived solidarity of the candidate was measured using multiple traits, including friendliness, niceness, warmth, likeability, trustworthiness, sincerity, honesty, and good-naturedness. These evaluations were typically measured via 5- to 7-point Likert -type scales (ranging from completely disagree to completely agree) or 7-point semantic differential scales (e.g., cold—warm, unlikable—likable ).
Linguistic cue	This moderator examined whether the strength of accent bias in hiring differed according to whether candidates spoke with regional versus foreign accents. We considered “regional” accents to include native variations of a language due to dialects (e.g., Northern American-accented English versus Southern American-accented English) or ethnic differences (e.g., standard American-accented English vs. African American English varieties) within a country as well as native variations of a language between countries (e.g., American-accented English vs. British-accented English). We considered “foreign” accents to include variations of a language as a result of the candidate having a different native language (e.g., a non-standard-accented candidate who speaks English with a Chinese accent due to their native language being Chinese).
Candidate gender	This moderator examined whether the strength of accent bias in hiring differed according to whether candidates were male or female.

Note. A standardized mean difference (Cohen’s d) in ratings on Accentedness, Comprehensibility, Status and Solidarity were calculated from the reported statistics (e.g., means and standard deviations for each candidate) from each individual study. Separate meta-analyses on the effect of candidate accent on these ratings were first conducted. Then, the effect sizes calculated from each individual study on these ratings were treated as continuous moderators for the effect of candidate accent on hiring decisions.

Table 2.

Primary Moderators of the Overall Effect of Candidate Accent on Hiring Decisions.

Moderator variable	ES	F(df₁, df₂)	Mean d	$β$	95% CIs		t₀
Moderator variable	ES	F(df₁, df₂)	Mean d	$β$	Lower	Upper	t₀
Continuous moderators
Accentedness	50	F(1, 48) = 0.33	—	–0.08	–0.35	0.19	—
Comprehensibility	52	F(1, 50) = 3.95	—	0.30	–0.00	0.60	—
Status	52	F(1, 50) = 72.65***	—	0.62	0.47	0.76	—
Solidarity	49	F(1, 47) = 5.93**	—	0.43	0.07	0.78	—
Categorical moderators
Job communication demands	40	F(1, 38) = 5.82*
Low	16		0.06	—	–0.49	0.61	0.21
High	24		0.62	—	0.15	1.09	2.67*
Linguistic cue	139	F(1, 137) = 8.08**
Foreign accent	85		0.62	—	0.36	0.88	4.77***
Regional accent	54		0.18	—	–0.13	0.49	1.16
Candidate gender	107	F(1, 105) = 39.64***
Male	75		0.11	—	–0.19	0.41	0.75
Female	32		1.13	—	0.80	1.46	6.76***

Note. For continuous moderators, we report $β$ (i.e., estimated regression coefficient), which conveys the increase in mean d (mean effect size for hiring bias favoring standard- over non-standard-accented candidates) for a unit increase in a given moderator variable. For categorical moderators, we report the mean d for each level of a given moderator variable. For each category of moderator variable, the reference variable is listed first. ES = number of effect sizes contributing to analysis; F(df₁, df₂) = omnibus test; CI = confidence interval; t₀ = difference in mean d for each level of categorical moderator from zero.

p < .05. **p < .01. ***p < .001.

Accentedness

Unsurprisingly, participants rated standard-accented candidates lower in accentedness relative to non-standard-accented candidates, d = −2.47; 95% CI [−3.25, −1.70], t(49) = 6.39, p < .001. However, accentedness was not a significant moderator of the effect of candidate accent on hiring decisions, $β$ = −0.08, 95% CI [−0.35, 0.19]. In other words, the difference between the standard- and non-standard-accented candidate in terms of accentedness was not a reliable predictor of the degree of hiring bias. This finding is surprising and casts doubt on the perspective that processing fluency concerns could exclusively explain the accent bias in hiring decisions.

Comprehensibility

Overall, participants rated standard-accented candidates higher in comprehensibility relative to non-standard-accented candidates, d = .98; 95% CI [0.49, 1.47], t(51) = 4.03, p < .001. However, meta-regression revealed that comprehensibility was not a significant moderator of the accent-related hiring bias, $β$ = 0.30, 95% CI [−0.00, 0.60]. Contrary to the processing fluency perspective on accent biases, this reveals that the difference between the standard- and non-standard-accented candidates in terms of comprehensibility was not a reliable predictor of the degree of hiring bias.

Job communication demands

Job communication demands significantly moderated the effect of candidate accents on hiring decisions. For jobs with high communication demands, a large and significant hiring bias in favor of standard-accented candidates emerged, d = 0.62, 95% CI [0.15, 1.09]. In contrast, for jobs with low communication demands, the mean effect size did not differ reliably from zero, d = 0.06, 95% CI [−0.49, 0.61]. Consistent with the processing fluency perspective, this finding supports the idea that non-standard accents could be seen as an impediment to communicate clearly and, by extension, to perform the job to a high standard. However, the absent moderating effect of comprehensibility on accent bias in hiring calls this interpretation into question. We discuss this in greater detail in the General Discussion.

Status

Participants rated standard-accented candidates higher in status relative to non-standard-accented candidates, d = 0.42; 95% CI [0.17, 0.67], t(51) = 3.41, p = .001. Furthermore, status was a significant moderator of the accent-based hiring bias, $β$ = 0.62, 95% CI [0.47, 0.76]. Consistent with the prejudice perspective, the more the standard-accented speaker was considered higher status relative to the non-standard-accented speaker, the more the standard-accented speaker was favored for the job over the non-standard-accented speaker. This finding supports the idea that non-standard accents activate low competence stereotypes (Gluszek & Dovidio, 2010) which, in turn, impedes non-standard candidates’ likelihood of being hired.

Solidarity

Overall, participants rated standard-accented candidates and non-standard-accented candidates equally in solidarity, d = −0.07; 95% CI [−0.21, 0.08], t(48) = 0.94, p = .35. One possible explanation for why candidates were judged equally in solidarity is because all effect sizes examining solidarity (except two) were based on data where between-subjects designs were used. Indeed, research suggests that compensation mechanisms (i.e., rating non-standard speakers higher in warmth to compensate for lower status ratings) are more likely to occur when within-subjects designs are used (e.g., Yzerbyt et al., 2005). Nevertheless, meta-regression demonstrated that the hiring bias was greater the more the standard speaker was favored in terms of solidarity, $β$ = 0.42, 95% CI [0.07, 0.78], although this appeared to depend on candidate gender (see below).

Linguistic cue

The linguistic cue was a significant moderator of the effect of candidate accent on hiring decisions. Standard-accented candidates were favored over foreign-accented candidates, d = 0.62, 95% CI [0.36, 0.88]. In contrast, standard-accented candidates were not favored any more than regional-accented candidates, d = 0.18, 95% CI [−0.13, 0.49]. Foreign accents in particular may activate negative stereotypes associated with non-native speakers (Gluszek & Dovidio, 2010).

We tested this possibility by examining the moderating effect of linguistic cue on the accent bias on ratings of status and solidarity. Interestingly, we found no evidence to suggest that status and solidarity ratings differed according to whether candidates spoke in regional versus foreign accents. Standard-accented candidates were rated higher in status relative to non-standard-accented candidates regardless of whether they differed by regional (d = 0.64, 95% CI [0.14, 1.15]) or foreign (d = 0.38, 95% CI [0.11, 0.65]) accent (see Table S1), and they were rated equally in solidarity regardless of whether candidates differed by regional (d = 0.08, 95% CI [−0.34, 0.49]) or foreign (d = −0.08, 95% CI [−0.24, 0.08]) accent (see Table S2). In both of these analyses, only a small number of effect sizes contributed to regional accents (i.e., eight and six, respectively); thus, caution should be taken in interpreting these findings. Nevertheless, in line with the prejudice perspective, these findings suggest that negative stereotypes—at least those related to status—are activated for non-standard-accented speakers, regardless of whether the accent was a regional or foreign variety.

Candidate gender

Candidate gender was a significant moderator of the effect of candidate accent on hiring decisions. When candidates were female, standard-accented candidates were strongly favored over non-standard-accented candidates, d = 1.13, 95% CI [0.80, 1.46]. In contrast, no accent bias in hiring was found for male candidates, d = 0.11, 95% CI [−0.19, 0.41]. This finding lends support to the prejudice perspective of accent biases: Specifically, female candidates’ voice may have signaled their belonging to a minority group (i.e., female), which in turn activated (negative) stereotypes.

To explore this possibility, we tested for the moderating effect of gender on the accent bias on ratings of status and solidarity. The effect of candidate accent on status ratings was not moderated by candidate gender (see Table S1)—that is, standard-accented candidates were rated higher in status compared with non-standard-accented candidates, regardless of whether they were male (d = 0.42, 95% CI [0.04, 0.80]) or female (d = 0.48, 95% CI [0.03, 0.93]). In contrast, the effect of candidate accent on solidarity ratings was significantly moderated by candidate gender (see Table S2). Participants rated standard-accented and non-standard-accented male candidates equally in solidarity, d = -0.11, 95% CI [−0.21, 0.00]. When both candidates were female, however, participants rated standard-accented candidates higher in solidarity relative to non-standard-accented candidates, d = 0.20, 95% CI [0.04, 0.37]. This finding suggests that female candidates that speak with a non-standard accent—but not male candidates—are downgraded on solidarity, and the degree to which they are downgraded on solidarity relative to the standard-accented candidate may hinder their chances of being hired. This finding supports the prejudice perspective that a non-standard accent (for women) activates a low warmth stereotype (Fuertes et al., 2012).

Secondary variables

Table 3 summarizes effect sizes and statistical output for each analysis of the secondary variables. Time exposed to candidate, hiring experience of participants, and job status were not significant moderators of the effect of candidate accent on hiring decisions. In addition, no moderators on article characteristics (year of publication and publication status) or sample characteristics (sample size, proportion males, and mean age) were significant.

Table 3.

Secondary Moderators of the Overall Effect of Candidate Accent on Hiring Decisions.

Moderator variable	ES	F(df₁, df₂)	Mean d	$β$	95% CIs		t₀
Moderator variable	ES	F(df₁, df₂)	Mean d	$β$	Lower	Upper	t₀
Continuous moderators
Time exposed to candidate	121	F(1, 119) = 0.01	—	0.00	–0.00	0.00	—
Year of publication	130	F(1, 128) = 0.45	—	0.01	–0.02	0.04	—
Sample size	139	F(1, 137) = 0.32	—	0.00	–0.00	0.00	—
Proportion males	139	F(1, 137) = 0.05	—	0.00	–0.02	0.02	—
Mean age	121	F(1, 119) = 0.26	—	–0.01	–0.03	0.02	—
Categorical moderators
Hiring experience	104	F(1, 102) = 0.49
Novice	83		0.46	—	0.18	0.73	3.30**
Expert	21		0.70	—	0.07	1.32	2.20*
Job status	35	F(1, 33) = 0.47
Low	18		0.24	—	–0.06	0.53	1.64
High	17		0.12	—	–0.18	0.42	0.79
Publication status	139	F(1, 137) = 2.46
Published	112		0.35	—	0.10	0.61	2.70**
Unpublished	27		0.72	—	0.33	1.12	3.69***

p < .05. **p < .01.

General Discussion

By providing the first statistical synthesis of the research on accent bias in hiring, we aimed to better understand why standard speakers appear to be consistently favored over non-standard speakers. This “why” question is critical as accent-based discrimination can often fly under the radar, resulting in a serious but largely unacknowledged problem in society (Gluszek & Dovidio, 2010).

Overall, results demonstrated a strong tendency for standard-accented candidates to be favored over non-standard-accented candidates. This effect was derived based on a large variety of accent contrasts. Job candidates presented to participants in the included studies spoke with five different native English accents (e.g., American-accented English), Mexican American-accented English, Arabic-accented English, five different European-accented English (e.g., French-accented English), three different Asian-accented English (e.g., Chinese-accented English), as well as standard German and nine different accents when speaking German (e.g., Russian-accented German). Thus, unlike individual studies which are often limited by a selection of specific accents, our meta-analysis provides the strongest evidence to date of the robustness and generalizability of accent-based biases in hiring.

Furthermore, findings from the moderator analyses provide insights into what psychological processes may drive this accent-based hiring bias. The strength of accent bias was greater when the communication demands of the job were high. On the surface, this finding appears consistent with the notion that non-standard accents are more cognitively effortful to process, which in turn, hinders job performance (the processing fluency explanation). However, this explanation is weakened by the finding that comprehensibility was not a significant moderator of hiring decisions—that is, while we found large differences in relative comprehensibility between the standard and non-standard-accented candidates overall, the degree to which candidates differed in their relative comprehensibility was not a reliable predictor of accent-based hiring biases.

So why did we find a significant effect of job communication demands but not comprehensibility? A possible explanation may be found when one examines how comprehensibility is measured. Specifically, all studies used a subjective measure of comprehensibility, asking participants to rate how easy candidates were to understand. In contrast to an objective measure (e.g., verbally repeating back or transcribing what an individual actually said), subjective comprehensibility measures are vulnerable to prejudice—that is, conscious or unconscious accent-based biases could result in exaggerated ratings of a candidate’s comprehensibility (Gluszek & Dovidio, 2010). Comprehensibility reasons alone may not influence accent-based hiring biases as people want to appear non-prejudiced. Comprehensibility-related reasons, however, could subsequently be used as a rational justification against hiring non-standard candidates when the job has high communication requirements (Hansen & Dovidio, 2016; Roessel et al., 2020). This interpretation is consistent with accounts of aversive prejudice; specifically, that negatively biased evaluations may be expressed, provided they can be rationalized as non-prejudiced (Brown, 2011; Roessel et al., 2020). Therefore, the disparity between the effects of job communication demands and comprehensibility on hiring decisions may be due to an individual being able to more easily rationalize a recruitment decision as job-focused (and non-prejudicial) when the job in question has high communication requirements (i.e., “ease of understanding is a necessary prerequisite for this job”) rather than just a comprehensibility reason alone (i.e., “they are difficult to understand”).

Other effects are also difficult to explain from a processing fluency perspective. One curious pattern (which we did not anticipate) was that perceptions of accentedness were also not reliably associated with hiring bias. There was a large difference between standard and non-standard speakers in terms of accentedness—this is effectively a manipulation check—but on average, the degree that participants registered the non-standard speaker as heavily accented was not related to the extent they displayed a hiring bias against that candidate. Whatever is driving the accent bias in hiring, it does not seem to be simply how much the speaker is perceived as “accented.”

Other moderator analyses support the notion that accent-based hiring biases may reflect prejudices that are driven by preexisting stereotypes. First, standard-accented candidates were rated higher in status overall relative to non-standard speakers. Furthermore, the degree to which candidates differed in status was a significant moderator of hiring decisions—that is, the more standard candidates were perceived as higher in status relative to non-standard candidates, the more standard candidates were favored for the job.

One advantage of conducting a meta-analysis was that we were able to test moderators that were not often examined in single studies. For example, the majority of studies either use exclusively male candidates or exclusively female candidates. On aggregate, evidence for an accent bias in hiring was found exclusively among female candidates. Why might this be? A deeper look into solidarity ratings revealed a striking finding: non-standard-accented female candidates (but not male candidates) were downgraded on solidarity, which may have impeded their likelihood of being hired. Therefore, one possible explanation may be due to the differing societal importance placed on women (more so than men) to appear warm (Ebert et al., 2014). Another possibility could be that language proficiency is stereotypically associated as female rather than male.

Similarly, some studies compared standard speakers to foreign-accented speakers and others compared standard speakers to regional-accented speakers. On aggregate, evidence for an accent bias in hiring was found exclusively among the foreign-accented candidates. The effect of accent on status and solidarity ratings, however, did not differ depending on whether the non-standard accent was a regional or foreign variety. Differences in factors other than status and solidarity evoked by foreign accents compared with regional accents, therefore, likely explain why the accent bias in hiring impacts foreign-accented candidates in particular. For example, regional accent varieties of one’s own native language may prompt in-group identification (Ryan, 1983) that buffers against lower status ratings, whereas the non-nativeness of foreign accents may signal “otherness” that—combined with lower status ratings—leads to hiring biases.

Taken together, these findings are difficult to explain from a processing fluency explanation. Rather, they appear more consistent with an aversive prejudice perspective: non-standard accents work against candidates when they already battle minority status. The meta-analysis offers the strongest evidence to date that voices that simultaneously signal multiple marginalized identities (e.g., non-standard accent, ethnic minority/out-group, female) are subjected to stronger discrimination in hiring contexts compared with voices that signal a single marginalized identity (e.g., non-standard accent). It raises the possibility that accents in themselves do not work against candidates, but hiring decisions may be influenced by broader (and perhaps unconscious) prejudices that rise to the surface based on the extent to which accents make salient one’s “otherness.”

Implications for Practice

Thus far, research on employment discrimination has largely focused on the role of visual cues, such as one’s skin color or gender, on hiring biases. This research has inspired the implementation of anti-discrimination policies specifically against race- and gender-based discrimination in companies worldwide (Nkomo & Kinahan, 2015). Indeed, many globalized companies today actively strive to increase racial and gender diversity in their teams (Barak, 2022). Efforts to increase diversity and respect for others based on the way one speaks, however, are virtually absent.

In light of our findings, we suggest several practical ways to mitigate accent-based hiring biases. First, recruiters should be trained to be aware of the potential biases that may influence their hiring decisions. In particular, recruiters should be made aware that non-standard accents can lead to unwarranted stereotypes related to status and solidarity. Indeed, participants do not display biased hiring decisions when they are informed that accents can unduly impact perceptions, and so should avoid allowing stereotypes to inform their evaluations (Roessel et al., 2019). Second, it is vital to reduce rationalizations around comprehensibility as a justification for biased hiring decisions (Roessel et al., 2020). Indeed, non-standard accents are most often downgraded on subjective ratings of comprehensibility, even when they are equally as intelligible as standard accents (Munro & Derwing, 1995). Moreover, within 1 min of listening, adults recognize words spoken by non-standard-accented speakers as quickly as words spoken by standard-accented speakers (Clarke & Garrett, 2004). It is essential, therefore, that recruiters are informed of the general ease of adapting to non-standard accents. Third (and relatedly), given that increased exposure to non-standard accents can reduce biases in comprehensibility ratings (Gass & Varonis, 1984), targeted training to increase recruiters’ familiarity with non-standard accents may facilitate increased objectivity in deciding the degree to which an individual’s spoken accent provides a genuine impediment to clear communication.

Limitations and Future Directions

Our meta-analysis included studies where the main point of difference between candidates was the accent they spoke, meaning the studies manipulated pronunciation-linked variables of the candidates (and not their appearance). However, because accent is a powerful cue to one’s ethnicity and nationality (Lippi-Green, 2012; Rakić et al., 2011), it is important to acknowledge that this meta-analysis may also be capturing biases against ethnic minorities or out-groups (as inferred via accent) more generally. As the overwhelming majority of research has studied accents by having participants listen to auditory stimuli in isolation, our meta-analysis was not able to test to what extent accent biases in hiring are impacted by more explicit cues to one’s race and/or ethnicity. The few studies that have examined both accent and race cues together, however, have consistently found that accent (over race) more strongly guides social categorization (e.g., Rakić et al., 2011) and evaluations (e.g., Hansen et al., 2017).

In addition, while this meta-analysis offers great scope across a large variety of accents, it is important to note that the accent varieties in the included studies were only examined in two languages: English and German. More research that spans a greater variety of languages sampled from different countries would provide further information on the generalizability of accent bias in hiring decisions.

Finally, our finding that hiring bias was only present for female candidates raises important questions on why some combinations of identities—and not others—appear to be more vulnerable to discrimination. In general, how the intersectionality of social categories and identities (e.g., accent, ethnicity, and gender) influences perceptions of speakers remain understudied (Dragojevic et al., 2020; but see Hansen et al., 2018; Rakić et al., 2020). Considering that intersectionality is tied to issues of power and status (Remedios & Akhtar, 2019) and that our findings reveal that accent is a powerful communicator of status, it is crucial to consider how accent intersects with other markers of identity. Such research is critical to understanding why some individuals are more or less likely to face discrimination when entering the workforce.

Conclusion

With a verbal interview being one of the most common assessment methods in recruitment, the way one speaks undeniably plays a critical role in determining hireability. Our meta-analytic synthesis revealed significant accent biases in hiring decisions that are difficult to argue exclusively from the processing fluency perspective. Vigilance to latent accent-based prejudices, therefore, must be kept at the forefront of our minds.

Supplemental Material

sj-docx-1-psp-10.1177_01461672221130595 – Supplemental material for Is Your Accent Right for the Job? A Meta-Analysis on Accent Bias in Hiring Decisions

Supplemental material, sj-docx-1-psp-10.1177_01461672221130595 for Is Your Accent Right for the Job? A Meta-Analysis on Accent Bias in Hiring Decisions by Jessica L. Spence, Matthew J. Hornsey, Eloise M. Stephenson and Kana Imuta in Personality and Social Psychology Bulletin

Footnotes

Acknowledgements

We would like to extend our gratitude to the researchers who kindly shared their data with us for this project.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jessica L. Spence

Supplemental Material

Supplemental material is available online with this article.

Notes

References

Ahmad

(2020). Do equal qualifications yield equal rewards for immigrants in the labour market? Work, Employment and Society, 34(5), 826–843. https://doi.org/10.1177/0950017020919670

Assink

Wibbelink

C. J.

(2016). Fitting three-level meta-analytic models in R: A step-by-step tutorial. The Quantitative Methods for Psychology, 12(3), 154–174. https://doi.org/10.20982/tqmp.12.3.p154

Barak

M. E. M.

(2022). Managing diversity: Toward a globally inclusive workplace. SAGE.

Baugh

(2000). Racial identification by speech. American Speech, 75(4), 362–364. https://doi.org/10.1215/00031283-75-4-362

Boreinstein

Hedges

Higgins

Rothstein

H. R.

(2005). Comprehensive meta-analysis (Version 2). Biostat.

Bradac

J. J.

(1990). Language attitudes and impression formation. In Giles

Robinson

W. P.

(Eds.), Handbook of language and social psychology (pp. 387–412). John Wiley.

*Bradac

J. J.

Wisegarver

(1984). Ascribed status, lexical diversity, and accent: Determinants of perceived status, solidarity, and control of speech style. Journal of Language and Social Psychology, 3(4), 239–255. https://doi.org/10.1177/0261927X8400300401

Brown

(2011). Prejudice: Its social psychology. John Wiley.

*Carlson

H. K.

McHenry

M. A.

(2006). Effect of accent and dialect on employability. Journal of Employment Counseling, 43(2), 70–83. https://doi.org/10.1002/j.2161-1920.2006.tb00008.x

10.

Cheung

M. W. L.

(2014). Modeling dependent effect sizes with three-level meta-analyses: A structural equation modeling approach. Psychological Methods, 19(2), 211–229. https://doi.org/10.1037/a0032968

11.

Cheung

M. W. L.

(2015). Meta-analysis: A structural equation modeling approach. John Wiley.

12.

Clarke

C. M.

Garrett

M. F.

(2004). Rapid adaptation to foreign-accented English. The Journal of the Acoustical Society of America, 116(6), 3647–3658. https://doi.org/10.1121/1.1815131

13.

*Cocchiara

F. K.

(2007). Sociolinguistic cues, perceived race, and employment selection outcomes: An exploration of the aversive racism framework [Doctoral dissertation, The University of Texas].

14.

*de Nijs

. (2019). The effects of non-native accents on hireability: A comparison of German-accented English, Spanish-accented English and American English in job interviews [Master’s thesis, Radboud University].

15.

*Deprez-Sims

A. S.

Morris

S. B.

(2010). Accents in the workplace: Their effects during a job interview. International Journal of Psychology, 45(6), 417–426. https://doi.org/10.1080/00207594.2010.499950

16.

*Deprez-Sims

A. S.

Morris

S. B.

(2013). The effect of non-native accents on the evaluation of applicants during an employment interview: The development of a path model. International Journal of Selection and Assessment, 21(4), 355–367. https://doi.org/10.1111/ijsa.12045

17.

Derwing

T. M.

Munro

M. J.

(2009). Putting accent in its place: Rethinking obstacles to communication. Language Teaching, 42(4), 476–490. https://doi.org/10.1017/S026144480800551X

18.

*de Souza

L. E. C.

Pereira

C. R.

Camino

de Lima

T. J. S.

Torres

A. R. R

. (2016). The legitimizing role of accent on discrimination against immigrants. European Journal of Social Psychology, 46(5), 609–620. https://doi.org/10.1002/ejsp.2216

19.

Dovidio

J. F.

Gaertner

S. L.

(1986). The aversive form of racism. In Dovidio

J. F.

Gaertner

S. L.

(Eds.), Prejudice, discrimination and racism (pp. 61–89). Academic Press.

20.

Dragojevic

Fasoli

Cramer

Rakić

(2020). Toward a century of language attitudes research: Looking back and moving forward. Journal of Language and Social Psychology, 40(1), 60–79. https://doi.org/10.1177/0261927X20966714

21.

Dragojevic

Giles

(2016). I don”t like you because you’re hard to understand: The role of processing fluency in the language attitudes process. Human Communication Research, 42(3), 396–420. https://doi.org/10.1111/hcre.12079

22.

Dragojevic

Giles

Beck

A.-C.

Tatum

N. T.

(2017). The fluency principle: Why foreign accent strength negatively biases language attitudes. Communication Monographs, 84(3), 385–405. https://doi.org/10.1080/03637751.2017.1322213

23.

Duval

Tweedie

(2000). Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56(2), 455–463. https://doi.org/10.1111/j.0006-341X.2000.00455.x

24.

Ebert

I. D.

Steffens

M. C.

Kroth

(2014). Warm, but maybe not so competent? Contemporary implicit stereotypes of women and men in Germany. Sex Roles, 70(9), 359–375. https://doi.org/10.1007/s11199-014-0369-5

25.

Edwards

(1982). Language attitudes and their implications among English speakers. In Ryan

E. B.

Giles

(Eds.), Attitudes toward language variation: Social and applied contexts (pp. 20–33). Edward Arnold.

26.

Egger

Smith

G. D.

Schneider

Minder

(1997). Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315(7109), 629–634. https://doi.org/10.1136/bmj.315.7109.629

27.

Fiske

S. T.

Cuddy

A. J.

Glick

(2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82(6), 878–902. https://doi.org/10.1037/0022-3514.82.6.878

28.

Floccia

Butler

Goslin

Ellis

(2009). Regional and foreign accent processing in English: Can listeners adapt? Journal of Psycholinguistic Research, 38(4), 379–412. https://doi.org/10.1007/s10936-008-9097-8

29.

Fuertes

J. N.

Gottdiener

W. H.

Martin

Gilbert

T. C.

Giles

(2012). A meta-analysis of the effects of speakers’ accents on interpersonal evaluations. European Journal of Social Psychology, 42(1), 120–133. https://doi.org/10.1002/ejsp.862

30.

Gass

Varonis

E. M.

(1984). The effect of familiarity on the comprehensibility of nonnative speech. Language Learning, 34(1), 65–87. https://doi.org/10.1111/j.1467-1770.1984.tb00996.x

31.

Giles

(1970). Evaluative reactions to accents. Educational Review, 22(3), 211–227. https://doi.org/10.1080/0013191700220301

32.

Giles

Watson

(2013). The social meanings of language, dialect, and accent: International perspectives on speech styles. Peter Lang.

33.

Gluszek

Dovidio

J. F.

(2010). The way they speak: A social psychological perspective on the stigma of nonnative accents in communication. Personality and Social Psychology Review, 14(2), 214–237. https://doi.org/10.1177/1088868309359288

34.

*Goatley-Soan

Baldwin

J. R.

(2018). Words apart: A study of attitudes toward varieties of South African English accents in a United States employment scenario. Journal of Language and Social Psychology, 37(6), 692–705. https://doi.org/10.1177/0261927X18800129

35.

Haddaway

N. R.

Collins

A. M.

Coughlin

Kirk

(2015). The role of google scholar in evidence reviews and its applicability to grey literature searching. PLoS ONE, 10(9), e0138237. https://doi.org/10.1371/journal.pone.0138237

36.

*Hancock

(n.d.). Linguistic bias in hiring: The effects of Asian accents in job interviews. Orange Coast College.

37.

Hansen

Dovidio

J. F.

(2016). Social dominance orientation, nonnative accents, and hiring recommendations. Cultural Diversity and Ethnic Minority Psychology, 22(4), 544–551. https://doi.org/10.1037/cdp0000101

38.

*Hansen

Rakić

Steffens

M. C.

(2014). When actions speak louder than words: Preventing discrimination of nonstandard speakers. Journal of Language and Social Psychology, 33(1), 68–77. https://doi.org/10.1177/0261927X13499761

39.

Hansen

Rakić

Steffens

M. C.

(2017). Competent and warm? How mismatching appearance and accent influence first impressions. Experimental Psychology, 64(1), 27–36. https://doi.org/10.1027/1618-3169/a000348

40.

*Hansen

Rakić

Steffens

M. C.

(2018). Foreign-looking native-accented people: More competent when first seen rather than heard. Social Psychological and Personality Science, 9(8), 1001–1009. https://doi.org/10.1177/0261927X13499761

41.

Hideg

Shen

Hancock

(2022). What is that I hear? An interdisciplinary review and research agenda for non-native accents in the workplace. Journal of Organizational Behavior, 43(2), 214–235. https://doi.org/10.1002/job.2591

42.

Hodson

Dovidio

J. F.

Gaertner

S. L.

(2002). Processes in racial discrimination: Differential weighting of conflicting information. Personality and Social Psychology Bulletin, 28(4), 460–471. https://doi.org/10.1177/0146167202287004

43.

*Hosoda

Nguyen

L. T.

Stone-Romero

(2012). The effect of Hispanic accents on employment decisions. Journal of Managerial Psychology. https://doi.org/10.1108/02683941211220162

44.

*Hosoda

Stone-Romero

(2010). The effects of foreign accents on employment-related decisions. Journal of Managerial Psychology, 25(2), 113–132. https://dx-doi-org.web.bisu.edu.cn/10.1108/02683941011019339

45.

*Huang

Frideger

Pearce

J. L.

(2013). Political skill: Explaining the effects of nonnative accent on managerial hiring and entrepreneurial investment decisions. Journal of Applied Psychology, 98(6), 1005–1017. https://doi.org/10.1037/a0034125

46.

Kinzler

K. D.

Dejesus

J. M.

(2013). Northern = smart and Southern = nice: The development of accent attitudes in the United States. Quarterly Journal of Experimental Psychology, 66(6), 1146–1158. https://doi.org/10.1080/17470218.2012.731695

47.

Knapp

Hartung

(2003). Improved tests for a random effects meta-regression with a single covariate. Statistics in Medicine, 22(17), 2693–2710. https://doi.org/10.1002/sim.1482

48.

S. J.

Judd

C. M.

Blair

I. V.

(2006). What the voice reveals: Within- and between-category stereotyping on the basis of voice. Personality and Social Psychology Bulletin, 32(6), 806–819. https://doi.org/10.1177/0146167206286627

49.

Lev-Ari

Keysar

(2010). Why don”t we believe non-native speakers? The influence of accent on credibility. Journal of Experimental Social Psychology, 46(6), 1093–1096. https://doi.org/10.1016/j.jesp.2010.05.025

50.

*Levon

Sharma

Watt

Cardoso

(2021). Accent bias and perceptions of professional competence in England. Journal of English Linguistics, 49(4), 355–388. https://doi.org/10.1177/00754242211046316

51.

*Levon

Sharma

Yang

(in press). Dynamic sociolinguistic processing: Real-time changes in judgements of speaker competence. Manuscript submitted for publication. Language Linguistic Society of America.

52.

Lindemann

(2003). Koreans, Chinese or Indians? Attitudes and ideologies about non-native English speakers in the United States. Journal of Sociolinguistics, 7(3), 348–364. https://doi.org/10.1111/1467-9481.00228

53.

Lippi-Green

(2012). English with an accent language, ideology, and discrimination in the United States (2nd ed.). Routledge.

54.

Lovakov

Agadullina

E. R.

(2021). Empirically derived guidelines for effect size interpretation in social psychology. European Journal of Social Psychology, 51(3), 485–504. https://doi.org/10.1002/ejsp.2752

55.

Moher

Liberati

Tetzlaff

Altman

D. G.

, & PRISMA Group. (2009). Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLOS Medicine, 6(7), e1000097. https://doi.org/10.1371/journal.pmed.1000097

56.

Moher

Shamseer

Clarke

Ghersi

Liberati

Petticrew

Shekelle

Stewart

L. A.

Group

P.-P.

(2015). Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic Reviews, 4(1), Article 1. https://doi.org/10.1186/2046-4053-4-1

57.

Munro

M. J.

Derwing

T. M.

(1995). Processing time, accent, and comprehensibility in the perception of native and foreign-accented speech. Language and Speech, 38(3), 289–306. https://doi.org/10.1177/002383099503800305

58.

Munro

M. J.

Derwing

T. M.

(1999). Foreign accent, comprehensibility, and intelligibility in the speech of second language learners. Language Learning, 49(s1), 285–310. https://doi.org/10.1111/0023-8333.49.s1.8

59.

*Nguyen

(2015). The interactive effects of accent, attire, and job status on employment-related decisions [Master’s thesis, San José State University].

60.

Nkomo

S. M.

Kinahan

(2015). Discrimination. In Wiley encyclopedia of management (pp. 1–3). https://doi.org/10.1002/9781118785317.weom110138

61.

Pantos

A. J.

Perkins

A. W.

(2013). Measuring implicit and explicit attitudes toward foreign accented speech. Journal of Language and Social Psychology, 32(1), 3–20. https://doi.org/10.1177/0261927X12463005

62.

*Poels

(2021). The influence of non-native accents on the hiring success [Bachelor’s thesis, Radboud University].

63.

*Rakić

Steffens

M. C.

Mummendey

(2011). Blinded by the accent! The minor role of looks in ethnic categorization. Journal of Personality and Social Psychology, 100(1), 16–29. https://doi.org/10.1037/a0021522

64.

Rakić

Steffens

M. C.

Sazegar

(2020). Do people remember what is prototypical? The role of accent–religion intersectionality for individual and category memory. Journal of Language and Social Psychology, 39(4), 476–494. https://doi.org/10.1177/0261927X20933330

65.

R Core Team. (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

66.

Remedios

J. D.

Akhtar

(2019). Intersectional approaches to the study of confronting prejudice. In Confronting prejudice and discrimination: The science of changing minds and behaviors (pp. 179–200). https://doi.org/10.1016/B978-0-12-814715-3.00010-2

67.

*Roessel

Schoel

Stahlberg

(2020). Modern notions of accent-ism: Findings, conceptualizations, and implications for interventions and research on nonnative accents. Journal of Language and Social Psychology, 39(1), 87–111. https://doi.org/10.1177/0261927X19884619

68.

*Roessel

Schoel

Zimmermann

Stahlberg

(2019). Shedding new light on the evaluation of accented speakers: Basic mechanisms behind nonnative listeners’ evaluations of nonnative accented job candidates. Journal of Language and Social Psychology, 38(1), 3–32. https://doi.org/10.1177/0261927X17747904

69.

Romero-Rivas

Morgan

Collier

(2021). Accentism on trial: Categorization/stereotyping and implicit biases predict harsher sentences for foreign-accented defendants. Journal of Language and Social Psychology, 41(2), 191–208. https://doi.org/10.1177/0261927X211022785

70.

Ryan

E. B.

(1983). Social psychological mechanisms underlying native speaker evaluations of non-native speech. Studies in Second Language Acquisition, 5(2), 148–159. https://doi.org/10.1017/S0272263100004824

71.

*Singer

Eder

G. S.

(1989). Effects of ethnicity, accent, and job status on selection decisions. International Journal of Psychology, 24(1–5), 13–34. https://doi.org/10.1080/00207594.1989.10600029

72.

Tabachnik

B. G.

Fidell

L. S.

(2013). Using multivariate statistics (6th ed.). Allyn & Bacon.

73.

*Timming

A. R.

(2017). The effect of foreign accent on employability: A study of the aural dimensions of aesthetic labour in customer-facing and non-customer-facing jobs. Work, Employment and Society, 31(3), 409–428. https://doi.org/10.1177/0950017016630260

74.

*Ulmer

M. A.

(1992). The effects of regional dialects on impression during interviews [Master’s thesis, Mississippi State University].

75.

United Nations Department of Economic and Social Affairs. (2019). International migration report 2019. https://www.un.org/en/development/desa/population/migration/publications/migrationreport/docs/InternationalMigration2019_Report.pdf

76.

*van de Wouw

(2020). The effects on non-native accented English on employability [Bachelor’s thesis, Radboud University].

77.

Viechtbauer

(2015). Meta-analysis package for R. https://cran.r-project.org/web/packages/metafor/metafor.pdf

78.

Viechtbauer

(2020). Assembling data for a meta-analysis of standardized mean differences. http://www.metafor-project.org/doku.php/tips:assembling_data_smd

79.

Wittenbrink

Schwarz

(2007). Implicit measures of attitudes. Guilford Press.

80.

Yakushko

Backhaus

Watson

Ngaruiya

Gonzalez

(2008). Career development concerns of recent immigrants and refugees. Journal of Career Development, 34(4), 362–396. https://doi.org/10.1177/0894845308316292

81.

*Yang

(2012). Does having a foreign accent affect men and women differently? Effect of foreign accent and gender on employment decisions and negotiations. Worcester Polytechnic Institute.

82.

*Yang

(in press). Testing the efficacy of anti-accent bias interventions [Unpublished data].

83.

Yzerbyt

Provost

Corneille

(2005). Not competent but warm. . . Really? Compensatory stereotypes in the French-speaking world. Group Processes & Intergroup Relations, 8(3), 291–308. https://doi.org/10.1177/1368430205053944

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB