The Processing Of Political Metaphors In Intercultural Contexts: The Role Of L2 Pronunciation In Metaphor Comprehension

In an increasingly globalized world, global leaders frequently deliver political speeches in English as a second language (L2) before diverse international audiences. Within political discourse, conceptual metaphors act as core cognitive mechanisms that simplify abstract socio-political issues and frame public perception by mapping familiar source domains onto complex target realities (Chahbane & Zrizi, 2023; Kövecses, 2008, 2016; Mio, 2009; Otieno et al., 2016). Processing these complex rhetorical figures requires substantial cognitive resources from listeners to decode linguistic inputs, manage narrative immersion, and construct social meaning (Luu, 2026; Otieno et al., 2016). However, the real-world communication readiness and rhetorical impact of these metaphors are heavily mediated by a fundamental phonological factor: the speaker's pronunciation. Pronunciation accuracy and accent patterns function as the primary cognitive filter, directly dictating how smoothly auditory acoustic features are mapped onto semantic networks in high-stakes communicative settings (Luu & Nguyen, 2026; Luu et al., 2025).

While classical studies grounded in Conceptual Metaphor Theory (CMT) emphasize the structural and ideological power of metaphors in political rhetoric (Chahbane & Zrizi, 2023; Mio, 2009; Zinken, 2003), existing research remains largely disconnected from the phonological constraints of real-time speech processing. Traditional literature heavily prioritizes textual analyses, assuming a uniform standard of delivery (Chahbane & Zrizi, 2023; Otieno et al., 2016). Conversely, psycholinguistic studies demonstrate that non-native pronunciation errors significantly impair intelligibility, elevate cognitive load, and trigger negative affective evaluations regarding a speaker’s authority (Johnson & Rosano, 2008; Luu & Nguyen, 2026; Luu et al., 2025). Despite separate advancements in these fields, a significant scholarly gap persists regarding how non-native pronunciation interacts with the cognitive processing of conceptual metaphors. When a politician speaks with a distinct foreign accent, the listener’s cognitive system faces a dual challenge: parsing acoustic deviations while simultaneously decoding abstract conceptual mappings (Luu & Nguyen, 2026; Otieno et al., 2016). This acoustic friction may create a cognitive bottleneck, exhausting the mental resources needed for semantic integration and lowering metaphorical awareness, which is vital for long-term information retention and vocabulary engagement (Pourdana et al., 2014).

Furthermore, the extent of this cognitive disruption is likely moderated by the cultural nature of the metaphors embedded within the discourse. CMT posits that conceptual metaphors are shaped by cultural contexts, separating into universal bodily orientations and highly specific intertextual frameworks (Kövecses, 2008, 2016; Zinken, 2003). While universal metaphors require less cognitive effort, culture-specific or context-sensitive metaphors demand that listeners navigate shared cultural knowledge and complex pragmatic markers, mirroring how communities negotiate digital slang or identity structures in mediated spaces (Luu et al., 2026). When delivered with non-native pronunciation, the cognitive fluency required to decode these culturally foreign structures may drop sharply, causing information loss or communication anxiety (Luu, 2026; Pourdana et al., 2014). For intercultural listeners navigating vocabulary learning challenges and varying proficiency levels (Johnson & Rosano, 2008; Luu & Le, 2026), this interaction could completely disrupt the persuasion loops that political actors rely on to build shared beliefs (Otieno et al., 2016).

To address these critical intersections, this study utilizes a controlled laboratory experiment using audio recordings to examine the real-time cognitive mechanisms underlying metaphor comprehension in political discourse. By systematically manipulating pronunciation accuracy (native standard pronunciation vs. non-native accented speech) and metaphor type (universal vs. culture-specific), this research provides empirical evidence of how phonological features influence semantic processing and message retention among intercultural listeners. Quantifying these effects through reaction times and recall metrics expands the scope of CMT to incorporate real-time auditory constraints (Kövecses, 2008; Mio, 2009). Ultimately, these findings offer practical communication strategies for diplomats navigating international media, highlighting that pronunciation mastery is a crucial asset for preserving the persuasive integrity of political messages in cross-cultural environments.

RQ1: Does a politician’s non-native pronunciation significantly hinder or delay the cognitive processing time of conceptual metaphors in political speeches among intercultural listeners?

H1: Listeners exposed to a foreign accent (Group B) will exhibit significantly longer reaction times when decoding the conceptual meaning of political metaphors compared to those exposed to standard native pronunciation (Group A).

RQ2: How does the interaction between pronunciation accuracy and metaphor type (universal vs. culture-specific) affect intercultural listeners' retention of political messages?

H2: Political message retention scores will be significantly lower in the foreign accent condition (Group B) than in the native condition (Group A), with this performance decrement being most pronounced when processing culture-specific metaphors.

METHODOLOGY

Research design

This study employs a quantitative, laboratory-based experimental design to investigate the causal effects of pronunciation accuracy and metaphor type on the real-time cognitive processing and long-term retention of political discourse. Specifically, a 2×2 mixed factorial design was deployed to examine these linguistic and cognitive intersections. The between-subjects factor was pronunciation accuracy, which consisted of two distinct auditory conditions: Standard Native English Pronunciation (Group A) and Non-Native Accented English Speech (Group B).

Meanwhile, the within-subjects factor was metaphor type, which systematically varied across two dimensions including Universal Metaphors, which are fundamentally grounded in shared human embodied experiences, and Culture-Specific Metaphors, which remain inherently dependent on regional, context-bound, or historical frameworks. By using this design, the study controls for individual learner differences while isolating how variations in auditory delivery influence semantic decoding and subsequent information mapping.

Research context and participants

The empirical investigation was conducted at a prominent public university located in Jakarta, Indonesia. This specific geopolitical and educational context offers an ideal research environment due to the expanding role of English as a primary medium for international diplomacy, socio-political discourse, and academic training within the ASEAN region. To ensure a highly homogeneous and qualified sample, a purposive sampling technique was utilized to recruit candidates from the university's student body based on four strict inclusion criteria. First, regarding nationality and linguistic background, all participants had to be Indonesian nationals who spoke Indonesian (Bahasa Indonesia) as their native language (L1). Second, in terms of academic status, participants had to be active, full-time undergraduate students majoring in English Education or English Literature to guarantee baseline linguistic readiness and familiarity with extended listening tasks. Third, for the English proficiency level, mastery at a certified Upper-Intermediate level (B2 according to the Common European Framework of Reference for Languages, CEFR) was required and verified via official institutional TOEFL or IELTS equivalent scores, ensuring that any observed processing delays were due to phonological friction rather than fundamental vocabulary or grammar deficiencies. Finally, participants had to meet strict cognitive and auditory health criteria, which required self-reported normal or corrected-to-normal hearing and vision, with no documented cognitive or learning disabilities.

In total, a sample of N = 106 undergraduate students successfully met all criteria, completed the experimental procedure, and were randomly assigned to one of the two between-subjects conditions, resulting in an equal distribution of n = 53 participants per group. This demographic distribution was balanced identically across the two experimental groups, with Group A experiencing native standard pronunciation and Group B interacting with the non-native accented condition.

Table 1. Demographic Characteristics of the Participants (N = 106)

INSTRUMENTS

Auditory Stimulus Material

The instrumentation architecture comprised three primary components specifically designed to deliver controlled stimulus manipulation, capture temporal cognitive data, and assess explicit information retention. The first instrument was the Auditory Stimulus Material, which consisted of a curated, high-stakes political speech of approximately 800 words focusing on global socio-economic alignment and national resilience. The text embedded exactly 20 conceptual metaphors, evenly divided into 10 Universal Metaphors, such as building a bridge to economic prosperity based on the structural mapping AN ORGANIZATIONAL PLAN IS A STRUCTURE, and 10 Culture-Specific Metaphors that required specific historical, socio-pragmatic, or regional contextual knowledge to be fully comprehended. For Condition A, the text was recorded by a professional male voice actor speaking standard General American English to represent native speech. For Condition B, the exact same text was recorded by a non-native English-speaking political figure exhibiting marked segmental and suprasegmental deviations, including L1 phonetic transfers, shifting syllable-timed rhythms, and altered lexical stress patterns. Both auditory recordings were calibrated to an identical average speech rate of 140 words per minute and normalized to a uniform amplitude of 70 dB to eliminate confounding acoustic variables.

Metaphor Cognitive Processing Test

The second instrument was the Metaphor Cognitive Processing Test, which measured real-time cognitive processing speed in milliseconds to evaluate the immediate processing latency required to decode the target metaphors. Administered via dedicated experimental software, the tool displayed the text of the speech scrolling smoothly across a computer monitor while the corresponding audio track was delivered to participants through high-definition headphones. Immediately following the vocalization of each of the 20 target metaphors, the audio playback automatically paused, and a binary semantic verification prompt appeared instantly on the screen, asking questions such as “Does this phrase imply collaborative unity?”. Participants were instructed to respond by pressing designated keyboard inputs as quickly and accurately as possible, with the letter "Y" corresponding to a positive verification and "N" representing a negative evaluation. The software recorded the exact latency between the visual prompt onset and the physical keypress, providing a precise measure of cognitive processing load under different phonological conditions.

Political Message Retention Test (PMRT)

The third instrument was the Political Message Retention Test (PMRT), an explicit, post-exposure comprehension assessment administered immediately after the listening task to evaluate long-term informational recall. The PMRT is a 10-item instrument scored out of a maximum of 10 points, where each item is worth exactly 1 point and no partial credit is awarded. The test is structurally split into two symmetrical sub-sections, where Section 1 contains 5 items evaluating explicit information retention tied to sentences utilizing universal conceptual mappings, and Section 2 contains 5 items assessing recall for sentences embedding culture-specific frameworks. Each item utilizes a structured, short-answer cued-recall format graded against a strict objective rubric. A score of 1.0 point is awarded if the response accurately reproduces the core semantic argument, explicit facts, or policy directions framed by the target metaphor while utilizing correct vocabulary matching the text. Conversely, a score of 0.0 point is assigned if the response is completely omitted, displays critical factual distortions, contains severe lexical ambiguities that alter the intended political meaning, or demonstrates a total failure to recall the core message.

DATA COLLECTION AND ANALYSIS

Experimental sessions were conducted individually in a quiet, isolated university laboratory using standard high-definition headphones and desktop computers. Prior to data collection, institutional ethics approval was formally secured from the university review board to protect participant welfare. All individuals received comprehensive information sheets outlining the study's precise scope, voluntary participation clauses, potential risks, and educational benefits, and subsequently provided written informed consent before commencing the task. Anonymity was strictly maintained throughout the project by assigning unique alphanumeric codes to all participants, and all digital records were stored securely on an encrypted drive accessible only to the primary research team.

Quantitatively, the reaction time data for the first research question and the political message retention scores from the PMRT for the second research question were input into SPSS 28.0 for statistical computation. Descriptive statistics were first calculated to establish baseline distribution profiles and check for normality. To evaluate Hypotheses 1 and 2, a series of mixed-design Analyses of Variance (ANOVAs) and independent-samples t-tests were computed to examine the main effects of pronunciation accuracy and the interaction effects across the two metaphor types, with the statistical significance threshold established at an alpha level of alpha = .05.

FINDINGS

Statistical Analysis for RQ1 and Hypothesis 1

To address the first research question and evaluate Hypothesis 1, which predicted that a foreign accent would significantly prolong cognitive processing times, the mean reaction times for semantic verification were computed for both experimental groups. The descriptive data revealed a clear disparity in processing latency between the two phonological conditions. Group A, the cohort exposed to standard native English pronunciation, demonstrated a faster and more efficient cognitive processing profile, maintaining an overall mean reaction time of 1245.60 milliseconds with a standard deviation of 112.40 milliseconds. In contrast, Group B, the cohort exposed to the non-native accented speech, exhibited an elevated processing latency, yielding a substantially higher mean reaction time of 1582.30 milliseconds alongside a standard deviation of 134.80 milliseconds. This descriptive divergence indicates that the acoustic deviations present in the accented delivery forced the intercultural listeners to expend more time decoding the auditory input before they could successfully evaluate the underlying conceptual metaphors.

To confirm whether this observed difference was statistically meaningful, an independent-samples t-test was calculated to compare the reaction times across the two groups. The inferential analysis revealed a highly significant effect of pronunciation accuracy on cognitive processing speed, t(104) = -14.01, p < .001. The magnitude of this difference was exceptionally large, as indicated by Cohen’s d = 2.72, which confirmed that the presence of a non-native accent exerted an intense disruptive influence on the real-time parsing of rhetorical language. Because the mean processing time for Group B was significantly elongated compared to Group A, the statistical evidence provides robust and unequivocal support for Hypothesis 1. This outcome empirically confirms that a politician's non-native pronunciation functions as an acoustic barrier that hinders and delays real-time semantic integration among upper-intermediate L2 listeners. These statistical comparisons for processing times are presented comprehensively in Table 2.

Table 2. Independent-Samples t-Test Results for Metaphor Cognitive Processing Time (Reaction Time in ms)

Statistical Analysis for RQ2 and Hypothesis 2

The second research question and Hypothesis 2 focused on the interactive effects of pronunciation accuracy and metaphor type on long-term political message retention. Total scores on the Political Message Retention Test (PMRT) were analyzed out of a maximum of 10 points. Group A achieved an overall mean retention score of 8.24 (SD = 0.85), while Group B scored noticeably lower with an overall mean of 5.32 (SD = 1.12). When broken down by sub-sections, Group A demonstrated high retention for both universal metaphors (M = 4.42, SD = 0.53) and culture-specific metaphors (M = 3.82, SD = 0.62). Conversely, Group B showed a modest drop in retention for universal metaphors (M = 3.51, SD = 0.69) but experienced a steep performance decline when processing culture-specific metaphors, averaging a low mean score of only 1.81 (SD = 0.74) out of a possible 5 points. The comprehensive descriptive metrics for these variables are structured in Table 3.

Table 3. Descriptive Statistics for Political Message Retention Test (PMRT) Scores

To thoroughly examine the main and interaction effects of these variables, a 2×2 mixed-design Analysis of Variance (ANOVA) was computed. The results yielded a significant main effect for the between-subjects factor of pronunciation accuracy, F(1, 104) = 214.56, p < .001, η2 = .67, confirming that an accented delivery significantly reduces general message recall. A significant main effect was also observed for the within-subjects factor of metaphor type, F(1, 104) = 158.42, p < .001, η2 = .60, indicating that culture-specific metaphors are inherently more difficult to retain than universal ones across both groups. Crucially, the ANOVA revealed a highly significant two-way interaction effect between pronunciation accuracy and metaphor type, F(1, 104) = 42.18, p < .001, η2 = .29. This statistical interaction confirms that the negative impact of a foreign accent was not uniform; instead, the performance decrement was significantly compounded when the non-native speech co-occurred with culturally specific conceptual frameworks. The complete mixed-design ANOVA parameters are documented in Table 4.

Table 4. Two-Way Mixed-Design ANOVA Results for Political Message Retention