Introduction
English functions as a lingua franca in a large share of international business, technology, healthcare, and academic settings, meaning that colleagues from different first-language backgrounds routinely conduct meetings, negotiations, and client interactions in a shared language that is not the first language of many participants involved [14][15][16]. For non-native speakers who reach a professional workplace, oral English proficiency is rarely a fixed, binary attribute. It interacts with anxiety, listener perception, and organizational context in ways that shape how competence is judged and how careers unfold. Research on this topic draws on two overlapping bodies of work: the psychological study of language anxiety, developed primarily in educational settings, and a smaller but growing literature on how non-native oral English functions specifically in professional and organizational contexts.
Prevalence of Language Anxiety in Professional and Workplace Contexts
Language anxiety, as a construct, was first formalized by Horwitz, Horwitz, and Cope, who introduced the Foreign Language Classroom Anxiety Scale (FLCAS) and conceptualized foreign language anxiety as a distinct, situation-specific form of anxiety composed of three interrelated components: communication apprehension, test anxiety, and fear of negative evaluation [1]. Although the FLCAS was developed and validated in classroom settings, its underlying components, particularly communication apprehension and fear of negative evaluation, describe emotional states that are not confined to formal instruction and that plausibly extend to any context in which a person must produce a non-native language under social evaluation, including the workplace.
MacIntyre and Gardner's foundational work on the cognitive effects of language anxiety helps explain why this matters functionally rather than only emotionally. Their research proposed that anxiety interferes with second-language performance at multiple stages of cognitive processing, including how information is taken in, processed, and retrieved for output, and that anxious learners tend to have reduced working-memory capacity available for language tasks compared with less anxious learners performing the same task [2]. Under this account, a professional who experiences elevated anxiety while speaking English in a meeting or presentation is not simply experiencing discomfort; the anxiety itself can measurably degrade the speaker's access to vocabulary, grammar, and fluent delivery in the moment.
Jean-Marc Dewaele's research extended language anxiety study specifically into multilingual populations, a population that overlaps substantially with non-native English-speaking professionals. In a study of communicative anxiety in second- and third-language production, Dewaele examined how psychological and sociodemographic factors relate to anxiety levels among multilingual speakers, finding that variables such as self-perceived proficiency and frequency of language use were associated with reported anxiety, while some background variables commonly assumed to matter were weaker predictors [3]. A later study extended this work to adult multilingual learners and examined how sociobiographical and situational factors relate to both general communicative anxiety and foreign language anxiety specifically, reinforcing that anxiety is not a fixed trait uniformly experienced across every language a multilingual person speaks, but instead varies with factors such as how often the language is used and in what context [4]. This is directly relevant to professional settings, where the frequency, formality, and stakes of English use vary considerably by role, industry, and seniority.
More recent research by Dewaele and MacIntyre examined the relationship between language anxiety and language enjoyment in parallel, finding that the two emotional states are not simple opposites and can coexist within the same learner or speaker, with each associated with a partially distinct set of individual and contextual factors [5]. This complicates any simple assumption that reducing anxiety alone is sufficient to produce confident, effective professional communication; positive engagement with the language appears to be a related but separable target.
Communication apprehension research conducted independently of second-language contexts, notably James McCroskey's foundational work on trait-like reluctance to engage in communication, provides an additional theoretical lens relevant to professional oral English, describing communication apprehension as a fear or anxiety associated with either real or anticipated communication that can suppress participation regardless of a speaker's underlying competence [17]. Combined with the second-language anxiety literature, this suggests that professionals speaking a non-native language at work may experience a compounding effect: general apprehension about speaking up, layered with anxiety specific to using a non-native language under professional scrutiny.
Career Advancement, Meeting Participation, and Perceived Competence
Beyond individual experience, language proficiency has measurable associations with professional and economic outcomes. In an analysis of international, multi-country data, Chiswick and Miller found a robust positive relationship between destination-language proficiency and immigrant earnings, controlling for other forms of human capital such as education and work experience, framing host-language skill as a significant, quantifiable labor-market asset in its own right rather than a secondary characteristic [19].
Organizational research adds detail to how this plays out inside workplaces. Tsedal Neeley's study of a Japanese multinational corporation that mandated English as its official corporate language documented how such a language shift can reorder status and influence within meetings: employees who were fluent in English, including some in relatively junior roles, gained outsized voice and visibility in discussions, while employees with strong technical expertise but limited English fluency reported anxiety, frustration, and reduced influence in the same settings, independent of their underlying job competence [14]. This body of work illustrates a specific mechanism connecting language proficiency to professional standing: meeting participation itself functions as a visible proxy for competence, and reduced participation due to language-related anxiety can be misread by colleagues as reduced capability.
Case-study research on English used as a lingua franca in Nordic corporate mergers similarly found that employees' confidence and comfort with English, distinct from their formal grammatical proficiency, shaped how actively they participated in cross-organizational communication after a merger, with English functioning less as a marker of native-like correctness and more as a practical, shared professional code [16]. Related research on lingua franca email communication within a multinational company found that employees adapted informally to workplace English, prioritizing clarity and functional efficiency over idiomatic or native-speaker norms [15], suggesting that professional linguistic communities often develop their own standards of adequacy that differ from classroom or native-speaker benchmarks.
Occupational research in regulated professions provides further evidence of how language requirements intersect with career access. A study examining English language testing requirements for internationally educated nurses seeking registration in the United Kingdom discussed the extent to which stringent language testing functions as a legitimate safeguard for patient safety versus a barrier that may operate as a form of institutionalized discrimination, situating this tension within nurses' broader difficulties integrating into the workplace [18]. This line of research highlights that professional gatekeeping mechanisms tied to language, such as licensing exams, can carry consequences for career access that extend well beyond communication effectiveness in a given interaction.
Consistent with these findings, survey research by Gluszek and Dovidio found that immigrants in the United States who perceived themselves as speaking English with a stronger non-native accent reported greater perceived discrimination and a lower sense of belonging, associations that plausibly extend into workplace experiences of exclusion or diminished standing even when not studied in an employment-specific sample [10]. A workplace-focused review by Foote similarly examined how accent perception operates specifically within professional settings, describing accent as a salient cue that colleagues, evaluators, and clients draw upon, often without deliberate intent, when forming judgments about a speaker's competence [25].
Accent, Disfluency, and Filler Words: Perceptual Effects on Listeners
A key contribution of Munro and Derwing's research is the finding that accentedness, intelligibility, and comprehensibility are separable, only partially correlated dimensions of second-language speech, rather than a single unified quality. Accentedness refers to how different a speaker's pronunciation sounds from a local variety, intelligibility refers to how much of the intended message a listener actually understands, and comprehensibility refers to how much subjective effort a listener perceives it takes to understand the speaker. Munro and Derwing demonstrated that speech can carry a strong, clearly detectable accent while remaining highly intelligible, meaning that accent alone does not necessarily predict how well a listener actually understands what was said [6]. Derwing and Munro extended this line of research across speakers from four different first-language backgrounds, again finding accentedness dissociable from comprehensibility and intelligibility ratings, with listener familiarity with a given accent also affecting judgments [7]. This distinction matters directly for workplace communication, since it implies that reducing a detectable accent is not the same intervention target as improving how well colleagues or clients actually understand a speaker.
Research on the acoustic features that drive listener judgments has focused increasing attention on suprasegmental, or prosodic, features of speech, meaning elements such as speech rate, stress patterns, and pausing, as distinct from segmental features, meaning the accuracy of individual vowel and consonant sounds. Kang, Rubin, and Pickering found that suprasegmental features were stronger predictors of listeners' judgments of a speaker's oral proficiency and perceived accentedness than segmental accuracy was, indicating that rhythm, pacing, and pausing patterns carry substantial weight in how competence is perceived, independent of grammatical or pronunciation accuracy [12]. This finding is directly relevant to filler words and disfluency pauses, since both are suprasegmental phenomena related to timing and flow rather than to sound accuracy. A related study by Kang examining ratings of international teaching assistants' oral performance found that both the prosodic features of a speaker's accentedness and characteristics of the raters themselves significantly influenced the resulting evaluations, indicating that assessments of non-native oral proficiency in professional and academic contexts are shaped by listener-side factors as well as speaker-side ones, and are not purely objective measurements of competence [13].
Perceptual bias extends beyond judgments of proficiency into judgments of the content itself. In an experimental study, Lev-Ari and Keysar had native-English-speaking listeners rate the truthfulness of factual statements read aloud by speakers with either a native or a non-native accent and found that statements delivered with a non-native accent were rated as less true, even though the content was identical, an effect the researchers attributed to the reduced processing fluency that comes with less familiar speech patterns rather than to deliberate distrust of the speaker [8]. Gluszek and Dovidio's broader theoretical review of accent stigma situates this kind of effect within a wider social-psychological framework, proposing that a non-native accent functions as a salient social category cue that can trigger bias in a manner comparable to other visible or audible markers of group membership, and reviewing evidence linking accented speech to negative social judgments across multiple domains [9]. Field research provides evidence that such perceptual dynamics can translate into concrete employment outcomes: a field experiment conducted in Germany by Schmaus and Kristen tested whether job applicants presenting with a foreign accent during initial employer contact experienced disadvantages in the early hiring process, finding evidence of accent-related disadvantage relative to applicants without a foreign accent, holding other applicant characteristics constant [11].
Evidence-Based Approaches to Professional Oral English Training
Taken together, the anxiety and perception literatures point toward two related but distinct training targets: reducing the anxiety response itself, and improving the specific communication patterns, such as pausing, pacing, and stress, that most strongly shape listener perception. Because MacIntyre and Gardner's model frames anxiety as interfering with cognitive processing at multiple stages, interventions that reduce situational anxiety during speaking practice, such as low-stakes rehearsal, repeated exposure, and structured practice with predictable formats, are consistent with the mechanism their model describes, even where a given intervention has not been tested as a formal clinical treatment for language anxiety specifically [2]. Similarly, because Horwitz, Horwitz, and Cope's model identifies fear of negative evaluation as a core component of language anxiety, training formats that reduce the sense of being evaluated, for instance through practice with peers, recorded self-review, or low-stakes simulated interactions rather than live high-stakes audiences, target a mechanism their framework identifies as central [1]. Dewaele and MacIntyre's finding that enjoyment and anxiety are separable further suggests that training approaches incorporating elements associated with enjoyment, rather than only anxiety reduction, may address a distinct and additive component of learner experience [5].
On the perception side, Munro and Derwing's comprehensibility research has been influential in shifting pronunciation instruction away from a goal of eliminating a foreign accent, which their research suggests is only weakly related to how well listeners understand a speaker, toward a goal of improving comprehensibility and intelligibility directly [6][7]. Given the suprasegmental findings from Kang, Rubin, and Pickering, this reframing has practical implications: instruction that addresses pausing patterns, speech rate, and sentence-level stress, rather than instruction focused narrowly on individual sound accuracy, targets the features most strongly associated with listener judgments of proficiency [12]. A multi-author review by O'Brien and colleagues, synthesizing the state of pronunciation research and technology, likewise recommends that instruction and assessment attend to both segmental and suprasegmental features together, and cautions that many existing tools remain better developed for segmental than for prosodic feedback, an imbalance the review identifies as a priority area for future development [22].
Technology-Assisted Interventions: Current Research and Outcomes
A body of empirical research has evaluated technology-assisted approaches to pronunciation and speaking development, most commonly under the umbrella of computer-assisted pronunciation training (CAPT), which typically uses automatic speech recognition (ASR) to provide learners with feedback on their spoken output. A meta-analysis by Mahdi and Al Khateeb, synthesizing 20 studies with a combined 1,014 participants, found a statistically significant, moderate-to-large positive overall effect of computer-assisted pronunciation training on foreign-language pronunciation outcomes, with larger effects observed for beginning and intermediate learners than for advanced learners, and for university-level learners compared with school-level learners. The authors also cautioned that many of the included studies were of low methodological quality, and recommended larger, better-controlled, multi-institutional research using standardized assessments before drawing strong conclusions about the size of the effect [20].
A more recent three-level meta-analysis by Hou and Min focused specifically on dialogue-based computer-assisted language learning systems, meaning tools that engage learners in interactive, conversation-style exchanges rather than isolated pronunciation drills, for second-language speaking development. Synthesizing 16 studies and 89 individual effect sizes, the analysis found a moderate overall positive effect on speaking development, and identified system type and interaction modality as significant moderators of outcomes, while learner proficiency level, educational setting, and several other factors examined did not significantly moderate results. The authors noted that the integration of generative artificial intelligence into dialogue-based systems remains an emerging and underexamined area, and called explicitly for further research on this point rather than treating it as an established finding [21].
Individual empirical studies have examined specific implementations of these technologies. Dai and Wu conducted a mixed-methods study of mobile-assisted pronunciation learning in which learners received feedback from peers, from automatic speech recognition, or from a combination of both, examining how the source of feedback related to learner engagement and outcomes [23]. Evers and Chen studied the effects of an automatic speech recognition system combined with peer feedback on pronunciation instruction for adult learners, contributing further evidence on how ASR-based tools function when combined with human feedback rather than used in isolation [24]. Across this literature, a consistent theme is that automatic speech recognition and related technologies show promising but still-developing evidence for supporting pronunciation and speaking gains, that effects are frequently moderated by how the technology is implemented and combined with human feedback, and that the research base, while growing, remains smaller and less mature than the classroom-based language anxiety literature described earlier in this article. The review by O'Brien and colleagues explicitly frames validating technology-assisted approaches against real-world communicative outcomes, such as comprehensibility in authentic interaction rather than performance on a narrow in-app task, as a priority for future research rather than a settled matter [22].
Limitations and Directions for Future Research
Several limitations characterize the current evidence base connecting language anxiety, accent perception, and professional outcomes. First, the foundational language anxiety instruments and much of the cognitive-processing research were developed and validated primarily in educational, not workplace, settings, and workplace-specific replications remain comparatively limited [1][2]. Second, most technology-assisted pronunciation and speaking research, including the meta-analyses summarized above, has been conducted with student populations in academic settings rather than with working professionals, meaning that findings about effect sizes and moderators may not generalize directly to workplace contexts, learner motivations, or the specific communicative demands of professional roles [20][21]. Third, much of the accent-perception literature relies on controlled experimental designs using brief speech samples or recordings, which may not fully capture how accent, disfluency, and other speech features interact with organizational, cultural, and interpersonal factors over the course of an ongoing professional relationship [8][9]. Longitudinal, workplace-embedded research connecting language anxiety, perceptual bias, and technology-assisted intervention outcomes to career trajectories over time remains a significant gap in the literature.
FAQs
What is language anxiety, and how is it typically measured in research?
Language anxiety refers to a situation-specific form of anxiety experienced when using a non-native language, distinct from general anxiety proneness. It was first formalized by Horwitz, Horwitz, and Cope through the Foreign Language Classroom Anxiety Scale, which models the construct as composed of communication apprehension, test anxiety, and fear of negative evaluation. Subsequent research, including work by MacIntyre and Gardner and by Dewaele, has extended and refined this construct, examining how it affects cognitive processing and how it varies across different languages, contexts, and individual speakers, including multilinguals using a language outside formal classroom settings.
Does a non-native accent reduce how well listeners understand what is said?
Not necessarily. Research by Munro and Derwing established that accentedness, meaning how different a speaker's pronunciation sounds from a local variety, is a separate and only partially related dimension from intelligibility, meaning how much of the message a listener actually understands, and comprehensibility, meaning how much effort a listener perceives it takes to understand. A speaker can have a strong, easily detectable accent while remaining highly intelligible. Separate research has found that listener judgments of a speaker's competence are often more strongly influenced by suprasegmental features, such as pausing and speech rate, and by social and cognitive biases related to accent itself, than by accent's actual effect on comprehension.
Can technology-based speech tools reduce speaking anxiety or build oral fluency for working professionals?
A growing body of research on computer-assisted pronunciation training and dialogue-based language learning systems has found generally positive, moderate effects on pronunciation and speaking development, based on meta-analyses synthesizing multiple studies. However, this research has been conducted predominantly with students in academic settings rather than with working professionals, and reviewers in the field have noted that the methodological quality of individual studies varies and that further research is needed to validate outcomes against real-world communicative performance rather than narrow task-based measures. The evidence is best characterized as promising but still developing, rather than conclusive for workplace populations specifically.