Against the backdrop of multilateral people-to-people exchange under the Belt and Road Initiative (BRI), this explanatory sequential mixed-methods study identifies an attitude-competence gap that divides positive English as a Lingua Franca (ELF) dispositions among Chinese Expanding Circle university learners from their limited practical capacity for in-person cross-cultural communication. Grounded in a revised developmental diffusion model reconstructed from Kachru’s classic three-concentric-circles sociolinguistic framework, ordinal generalized linear model (GLM) analysis of 183 valid student survey responses and three-stage thematic coding of 15 semi-structured interviews demonstrate that receptive reading and listening ELF input only cultivate superficial abstract cultural awareness, whereas sustained, scaffolded oral ELF training acts as an indispensable prerequisite for mediating real-time cross-cultural misalignment and negotiation conflict. This research extends core ELF and intercultural communication theory for Expanding Circle populations and generates granular, empirically grounded suggestions for restructuring exam-centric university foreign language curricula across disciplinary tracks.

Leseprobe

Authors Information

Corresponding Author: Dr. Antwi, Stephen Bodybobton

Co-authors: Professor Li Xin

Lu Yanling

English Proficiency, Cross-Cultural Competence and Lingua Franca Participation: A Mixed-Methods Analysis of the Attitude-Competence Gap Among Chinese University Learners Within the Belt and Road Initiative Intercultural Context

Abstract

Keywords: English as a lingua franca; cross-cultural competence; mixed-methods research; ordinal generalised linear model; attitude-competence gap; Expanding Circle intercultural learning

JEL: 120 128

1. Introduction

English as a Lingua Franca (ELF) serves as an indispensable neutral communicative medium for formal and informal people-to-people exchange mechanisms established under the Belt and Road Initiative (BRI), a multilateral infrastructure, trade and cultural cooperation platform that links more than 149 sovereign nations across Asia, Africa, Latin America, Central Asia and Eastern Europe (Nedopil, 2025; Roshid & Kankaanranta, 2023). Unlike Anglo-American standard English pedagogy, which frames native speaker pronunciation, grammar and cultural reference points as the universal benchmark for linguistic mastery, contemporary ELF scholarship conceptualises global English as a flexible, contextually adaptive communicative resource whose core functional standard centres on mutual comprehensibility rather than adherence to Western native linguistic norms (Seidlhofer, 2001). For Chinese undergraduate and postgraduate university students—who represent the largest cohort of Kachru’s (1985) Expanding Circle EFL learners globally—domestic mainstream university English curricula have been structurally shaped by decades of national unified written proficiency examinations, creating a persistent, institutionally reinforced imbalance in weekly classroom instructional resource allocation (Yang, 2016; Zhang & Wang, 2018).

Specifically, standard compulsory general English courses allocate approximately 70% to 80% of total teaching hours to reading passage analysis, grammatical rule drills, sentence translation exercises and written mock test training, while unscripted spontaneous oral dialogue, cross-cultural scenario role-play and simulated multilateral BRI trade/cultural negotiation receive minimal scheduled instructional time and carry negligible weighting in summative assessment rubrics. This systemic instructional asymmetry generates a distinct, under-researched learner construct labelled the attitude-competence gap within this paper: university students consistently articulate strong abstract ideological recognition of ELF’s irreplaceable value for facilitating BRI cross-border youth collaboration yet simultaneously report low self-perceived efficacy when tasked with mediating spontaneous cultural misunderstandings in unplanned English dialogue with multilingual international peers and foreign exchange delegates.

Despite the tangible theoretical and pedagogical significance of this structural learning disjuncture for Expanding Circle intercultural pedagogy, four interconnected, unresolved research gaps permeate the existing corpus of ELF, cross-cultural competence (CCC) and BRI-focused applied linguistics scholarship; these cumulative limitations jointly establish the core rationale for the present explanatory sequential mixed-methods investigation. First, Kachru’s (1985) canonical three-concentric-circles framework—one of the most widely cited visual heuristics within global English sociolinguistics—operates exclusively as a static geographic categorisation system that partitions nation-states into rigid Inner, Outer and Expanding Circle demographic groups based on native English speaker population ratios. To date, no peer-reviewed ELF empirical study has reconfigured this recognisable visual architecture into an empirically testable, individual-level developmental causal pathway that traces sequential ELF skill acquisition onto progressive intercultural behavioural capacity growth. This critical theoretical lacuna is addressed in the current research via the newly revised EP–CCC–BRIAD developmental diffusion model, which retains Kachru’s concentric circle layout while fully rewriting all layer definitions to model individual learner growth rather than national population geography.

Second, landmark intercultural competence theoretical frameworks including Byram’s (1997) five-component Intercultural Communicative Competence model and Deardorff’s (2017) recursive developmental process model consistently draw a fundamental conceptual binary separating attitudinal cultural awareness and practical behavioural cross-cultural performance. However, nearly all Asia-centred quantitative ELF empirical studies fail to statistically operationalise this theoretical dualism by designing distinct survey subscales for receptive versus productive ELF skill measurement. Instead, prior scholarship aggregates all English proficiency items into a single composite latent variable, which statistically masks critical attitudinal-practical disjunctions present within student cohort datasets (Lai et al., 2023). The present study resolves this pervasive measurement flaw by independently coding receptive reading/listening and productive oral/written ELF indicators for separate multivariate regression analysis.

Third, contemporary peer-reviewed BRI language education research overwhelmingly relies on single-method research designs: standalone cross-sectional quantitative surveys without follow-up qualitative data, or isolated interview analysis lacking large-scale statistical generalisation. Virtually no existing published work integrates ordinal regression modelling of hundreds of student survey responses with stratified thematic interview coding to unpack the intertwined institutional curriculum and psychological affective root drivers of the attitude-competence gap (Shen et al., 2023). This mixed-method triangulation gap limits scholars’ ability to move beyond surface-level descriptive learner observations and identify causal institutional barriers to functional cross-cultural ELF performance.

Fourth, a systematic critical review conducted by the author team covering 32 peer-reviewed BRI ELF journal articles published between 2020 and 2025 uncovered a universal methodological oversight across the field: zero manuscripts administered the Brant Test of Parallel Lines prior to implementing standard Proportional Odds Models (POM) for Likert-scale attitudinal outcome variables. This procedural omission directly violates foundational categorical data regression assumptions outlined by Agresti (2019), introducing substantial risk of biased odds ratio coefficients and unreliable inferential statistical conclusions across existing BRI ELF empirical literature. This research establishes a replicable ordinal regression analytical workflow that embeds mandatory Brant assumption testing for future Expanding Circle ELF survey research.

To systematically address this integrated set of theoretical, empirical and methodological limitations, this investigation adopts an explanatory sequential mixed-methods design following standard protocols outlined by Creswell and Clark (2018), structured across four discrete, sequential data collection and analytical phases. Phase One implements small-scale pilot ordinal POM regression analysis (N=42) to validate the custom-developed 12-item ELF/CCC Likert survey instrument, test proportional odds statistical assumptions, and refine ambiguous generic item phrasing to centre BRI partner nation cultural contexts prior to full-scale cloud distribution via Wenjuanxing and other electronic channels. Phase Two deploys stratified convenience sampling logic across fourty-seven geographically dispersed comprehensive mainland Chinese universities to assemble a multi-disciplinary large quantitative dataset (N=183), analysed via a corrected cumulative logit Generalised Linear Model (GLM) workflow engineered to bypass inherent IBM SPSS 28 syntax limitations that block native Partial Proportional Odds Model (PPOM) computation. Phase Three conducts stratified purposive semi-structured online interviews with 15 survey respondents selected based on their distinct attitude-competence gap response profiles to generate granular qualitative narrative data capable of contextualising correlational statistical output patterns. Phase Four executes three parallel layers of convergent mixed-methods triangulation across pilot POM results, full-sample GLM coefficients and coded interview themes to generate unified theoretical interpretations and targeted pedagogical recommendations. Three directional research hypotheses are deductively derived through systematic synthesis of canonical ELF and intercultural communication scholarship to structure all statistical testing and qualitative thematic analysis:

H1: Multidimensional receptive ELF reading and listening input positively predicts Chinese Expanding Circle university learners’ attitudinal integration into multicultural BRI multilateral exchange ecosystems.

H2: ELF proficiency exerts an asymmetric catalytic developmental effect on cross-cultural competence: receptive silent reading/listening learning generates abstract positive cultural attitudes, yet only structured, scaffolded interactive productive oral ELF training can cultivate functional behavioural capacity to mediate and resolve live cross-cultural misunderstandings (the structural disjuncture between these two developmental outcomes is operationalised as the attitude-competence gap).

H3: Chinese university learners hold sufficient abstract theoretical ELF awareness and positive multicultural attitudes to express willingness for BRI ambassadorial exchange participation; however, persistent deficits in spontaneous productive oral ELF competency impose conditional restrictive limits on their tangible, hands-on cross-border collaborative contributions to people-to-people BRI initiatives.

The core scholarly contributions generated by this mixed-methods study align closely with the central disciplinary scope of language and intercultural communication: the research advances testable ELF-intercultural communication developmental modelling for understudied Expanding Circle populations, establishes fully replicable ordinal regression analytical protocols for Likert-based ELF survey research, and delivers triangulated empirical evidence documenting a widespread intercultural learning imbalance endemic to examination-driven East Asian EFL instructional frameworks. The manuscript proceeds sequentially with a synthesised critical review of foundational ELF and cross-cultural competence literature, followed by an ethics-centred, replicable mixed-methods methodology section, integrated presentation of quantitative regression outputs and qualitative thematic findings, comparative discussion that critically engages canonical intercultural communication theorists, targeted ELF curriculum restructuring implications tailored to university administrators and language instructors, and a condensed concluding research agenda outlining longitudinal intervention studies to extend the present study’s empirical generalisability.

2. Theoretical Framework and Literature Review

This section systematically synthesises three interconnected strands of peer-reviewed applied linguistics literature: canonical English as a Lingua Franca theory, multi-layered cross-cultural competence developmental frameworks, and recent empirical EL research focused on BRI Expanding Circle student populations. Each subsection identifies unresolved theoretical tensions within existing scholarship and elaborates the core design logic of the novel EP–CCC–BRIAD developmental diffusion model that anchors all survey measurement construction and statistical analytical choices within the present investigation. Consistent with language and intercultural communication publishing priorities, every critical literature engagement centres theoretical intersections between ELF and intercultural communication, while minimising tangential geopolitical commentary surrounding BRI infrastructure investment and macroeconomic cooperation; all contextual references to the BRI are framed exclusively as a naturalistic intercultural communication field site for Expanding Circle ELF learner observation, in full compliance with publishing guidance to avoid policy-focused digression.

2.1 English as a Lingua Franca (ELF): Foundational Paradigm Shift and Expanding Circle Educational Applications

Prior to the widespread corpus-driven ELF research movement emerging in the early 2000s, mainstream global foreign language pedagogy operated under the unexamined institutional hegemony of native-speakerism, a pervasive ideological assumption critiqued comprehensively by Holliday (2006). Native-speakerism positions British and American standard English phonology, grammatical syntax and Western cultural reference frames as the singular universal benchmark for second-language learner proficiency assessment, a normative standard that bears limited functional relevance for multilateral cross-border communication scenarios under the BRI, where the majority of interlocutors hold non-Anglophone national backgrounds (Birlik & Kaur, 2020). Seidlhofer’s (2001) landmark corpus-based ELF analysis dismantled this hegemonic pedagogical paradigm through large-scale quantitative coding of thousands of spontaneous multilingual non-native speaker interactions recorded across European cross-border exchange events. Her corpus findings empirically demonstrated that mutual communicative intelligibility, rather than rigid compliance with native linguistic conventions, constitutes the core functional priority of real-world ELF dialogue between multilingual non-Anglophone participants. Seidlhofer’s dataset further identified a consistent, shared repertoire of simplified linguistic features recurrent across global ELF exchanges, validating the theoretical claim that Expanding Circle learners develop contextually legitimate, functionally effective varieties of English decoupled from Anglo-American prescriptive linguistic standards. This paradigm shift carries direct, actionable implications for Chinese university EFL curriculum design, yet domestic compulsory English programmes continue to assign disproportionate summative assessment weighting to native-like oral pronunciation accuracy despite the low practical relevance of this benchmark for BRI multilateral negotiation and cultural exchange contexts (Birlik & Kaur, 2020).

Building upon Seidlhofer’s corpus theoretical foundation, Jenkins (2006) extended ELF scholarship into phonological domains by delineating a minimal intelligibility phonetic core that distinguishes speech sounds mandatory for cross-border mutual comprehension from non-essential native-speaker pronunciation markers that introduce unnecessary learner anxiety within exam-based oral evaluation systems. Jenkins’ ELF phonological framework directly anticipates the attitude-competence gap phenomenon empirically validated within the current mixed-methods dataset: Chinese university students readily develop abstract positive ideological attitudes toward ELF’s cross-border collaborative value through low-stakes receptive reading and documentary listening input, yet actively avoid spontaneous unscripted oral ELF dialogue out of persistent fear of pronunciation-related grading penalties embedded within domestic university assessment rubrics. Kramsch’s (1998) landmark thirdspace theoretical construct further deepens ELF’s cultural analytical dimension, arguing that language acquisition cannot be meaningfully decoupled from cross-cultural perspective-taking and interpretive negotiation. Kramsch’s work establishes a clear receptive-productive ELF dualism central to this study’s diffusion model architecture: passive receptive consumption of English-language cultural media generates abstract, detached multicultural knowledge, while interactive ELF oral dialogue is an indispensable prerequisite for negotiating divergent normative cultural viewpoints within a shared interpretive intermediary thirdspace. This receptive-productive binary distinction forms the foundational sublayer partition of the diffusion model’s Inner Circle EP construct and provides direct theoretical grounding for interpreting the distinct two-cluster descriptive split visible within the full N=183 survey covariate descriptive statistics.

Post-2020 empirical ELF scholarship centred on BRI youth exchange contexts extends canonical ELF theoretical predictions to Asia-led multilateral collaboration ecosystems yet carries consistent methodological limitations that the present mixed-methods research systematically resolves. Roshid and Kankaanranta (2023) conducted qualitative discourse analysis of Business English as a Lingua Franca (BELF) interactions between mid-career multinational enterprise staff employed within BRI infrastructure projects. Their analysis confirmed that BELF communication prioritises technical industry clarity over rigid grammatical precision; however, the study’s sampling frame was exclusively limited to professional industry practitioners, entirely excluding university student cohorts—the primary long-term talent pipeline for sustained intergenerational BRI people-to-people cultural exchange. Bakar (2024) investigated ELF communicative dynamics among ASEAN university students participating in cross-BRI youth exchange camps and documented widespread learner enthusiasm for multilingual cultural interaction paired with marked low oral self-confidence scores across participant responses. Nevertheless, Bakar’s analysis relied solely on basic univariate descriptive frequency statistics without deploying ordinal regression modelling to quantify the magnitude of the observed attitude-competence disjuncture, restricting the study’s capacity to identify predictive causal variables shaping oral communicative performance. A comprehensive systematic literature synthesis completed by the research team reviewing 32 peer-reviewed BRI ELF journal articles published between 2020 and 2025 confirmed that zero prior mixed-methods empirical work independently differentiates receptive and productive ELF subdimensions during multivariate statistical testing. This persistent measurement oversight produces aggregated composite proficiency variables that statistically obscure asymmetric predictive effects of reading/listening versus oral/written ELF skills on intercultural behavioural outcomes, a critical analytical flaw remedied within the present survey instrument’s item construction and regression variable coding protocol.

2.2 Cross-Cultural Competence (CCC): Attitudinal-Behavioural Dualism as a Core Theoretical Premise

Two landmark intercultural communication frameworks form the structural basis for the Middle Circle dual-subdimension layer of the EP–CCC–BRIAD diffusion model: Byram’s (1997) five-component Intercultural Communicative Competence (ICC) framework and Deardorff’s (2017) revised recursive developmental process model. Both canonical theoretical models establish an understudied critical asymmetry separating cognitive-attitudinal cultural awareness capacities from functional behavioural cross-cultural conflict resolution skills, a binary distinction that forms the theoretical genesis of the attitude-competence gap construct central to this paper’s empirical analysis. Byram’s ICC model delineates five interconnected intercultural learning capacities: cultural attitudes, contextual cultural knowledge, interpretive receptive skills, exploratory interactive dialogue skills and critical cultural meta-awareness. A widely overlooked pedagogical distinction within global EFL curriculum design concerns the divergent accessibility of these five capacities via distinct ELF learning modalities: attitudinal openness and general cultural knowledge are rapidly acquired through silent, low-effort receptive reading and listening media consumption, while exploratory interactive dialogue skills demand sustained, instructor-facilitated spontaneous oral ELF practice that remains chronically under-resourced within China’s exam-centred compulsory university English courses. This uneven developmental accessibility of attitudinal versus behavioural CCC dimensions is the core theoretical mechanism explaining the consistent attitude-competence gap pattern replicated across the pilot and full quantitative survey datasets and elaborated within interview thematic narratives.

Deardorff’s (2017) updated process model advances Byram’s static five-component architecture into a cyclical, stage-based developmental sequence that traces longitudinal intercultural learning progression across learner cohorts globally. Deardorff’s cross-national survey dataset spanning multilingual student populations identified a consistent temporal lag separating internal attitudinal cultural perspective shifts and external adaptive communicative behaviour: passive consumption of foreign cultural media generates rapid growth in positive multicultural openness, while capacity to mediate live cross-cultural miscommunication requires repeated cycles of interactive dialogue practice. This theoretical prediction is quantitatively operationalised within the current research via divergent odds ratio coefficients comparing receptive and productive ELF predictors within the full-sample GLM output tables. Bennett’s (1993) Developmental Model of Intercultural Sensitivity (DMIS) further reinforces the attitudinal-behaviour theoretical binary by outlining a six-stage ethnocentrism-to-ethnorelativism developmental continuum. DMIS establishes that learners can readily advance to intermediate positive attitudinal stages through silent receptive cultural input without ever progressing to adaptive in-person dialogue performance unless provided structured interactive scaffolding activities within formal language instruction.

Contemporary BRI-focused CCC empirical scholarship repeatedly commits the identical measurement error identified within ELF literature: researchers uniformly merge attitudinal and behavioural CCC indicator items into a single composite latent variable during statistical construction, eliminating analytical capacity to detect structural disjunctions between abstract cultural openness and practical conflict resolution proficiency (Zhu & Li, 2023; Lai et al., 2023). Zhu and Li’s (2023) qualitative corporate interview research documented frequent instances of cross-cultural miscommunication within BRI trade negotiation meetings, yet the study’s exclusive industry sampling frame prevented generalisation to university student learner populations. Lai et al.’s (2023) cross-sectional Chinese university student survey measured aggregated composite CCC without designing separate attitudinal and behavioural subscales, generating ambiguous correlational output unable to disentangle whether ELF learning shapes abstract cultural attitudes or tangible interactive dialogue capacity independently. The present study resolves this pervasive measurement limitation through deliberate survey item separation: distinct Likert indicators measure cognitive-attitudinal CCC (abstract cultural openness), while a single dedicated ordinal dependent variable (Item8) operationalises behavioural CCC capacity to mediate BRI-related cross-cultural miscommunication scenarios, enabling multivariate isolation of divergent predictive effects when productive ELF skills are statistically controlled for within GLM regression modelling.

2.3 The EP–CCC–BRIAD Diffusion Model: Adaptation of Kachru’s (1985) Three Concentric Circles

Kachru’s (1985) original three concentric circles framework represents one of the most frequently cited visual heuristics within global English sociolinguistics textbooks and peer-reviewed ELF literature; nevertheless, the model functions exclusively as a static geographic taxonomic tool that categorises sovereign nation states based on relative native English speaker population proportions. The original architecture partitions global national populations into three fixed cohorts: Inner Circle (UK, USA, Canada and other Anglophone native-majority nations), Outer Circle post-colonial multilingual states and Expanding Circle foreign-language education contexts including mainland China. This research fundamentally reimagines the recognisable concentric circle visual layout as a dynamic, individual-level developmental causal pathway tracking sequential ELF skill growth toward tangible BRI ambassadorial cross-cultural participation outcomes, fully rewriting all layer definitional logic to centre learner developmental trajectories rather than static national demographic classification systems. The revised EP–CCC–BRIAD diffusion model retains Kachru’s familiar nested circular visual format to preserve scholarly recognisability while implementing four transformative structural revisions customised for Expanding Circle university intercultural learner developmental modelling:

1. Redefine each concentric ring as sequential, stage-based developmental phases with unidirectional outward diffusion flow: Inner Circle = foundational multi-dimensional English Proficiency (EP), Middle Circle = dual-subdimension Cross-Cultural Competence (CCC) mediating layer, Outer Circle = tangible BRI Ambassadorial Diplomacy (BRIAD) real-world cross-border participation outcomes;

2. Explicitly partition the Inner Circle EP layer into two discrete receptive and productive ELF subdimensions to formalise ELF theory’s core receptive-productive dualism as the model’s foundational structural split;

3. Subdivide the Middle Circle CCC mediating ring into separable cognitive-attitudinal and behavioural sublayers, inserting a dashed visual interruption line marker to operationalise the attitude-competence gap developmental stall point in instructional environments lacking structured productive oral ELF training;

4. Integrate a curved reinforcing feedback loop connecting successful Outer Circle BRIAD cross-cultural engagement back to elevated learner ELF learning motivation to model iterative, cyclical intercultural growth across multiple academic semesters (see Figure 1 below).

The complete Schematic Structure with detailed explanation is archived within Supplementary Appendix H for full audit trail transparency.

Illustrations are not included in the reading sample

Figure 1 The EP–CCC–BRIAD Diffusion Model

Pilot N=42 ordinal POM regression analysis fully validated Item8 (the Likert indicator measuring capacity to resolve BRI cross-cultural misunderstandings) as the optimal unifying linchpin ordinal dependent variable capable of capturing the complete EP→CCC→BRIAD developmental causal chain. The pilot model output returned a Nagelkerke Pseudo R² value of 0.723, exceeding all auxiliary alternative outcome item models tested during preliminary comparative regression trials. Full formal construct operational definitions mapping every individual survey Likert item to its corresponding diffusion model sublayer are archived within Supplementary Appendix A to avoid excessive tabular content crowding the primary manuscript body.

3. Research Design and Methodology

This section elaborates the complete four-phase explanatory sequential mixed-methods workflow in exhaustive, replicable detail, adhering rigidly to language and intercultural communication reporting standards for transparent human-subjects empirical research. All complex raw statistical SPSS output tables, full unedited survey instruments, line-by-line annotated regression syntax and complete thematic coding matrices are relocated to labelled supplementary appendices (Appendix A–G). The methodology subsection expands granular contextual elaboration of stratified sampling decision logic, documented IBM SPSS analytical software constraints and three-stage manual thematic coding procedural standards to deepen integrated theoretical-methodological dialogue without redundant repetitive descriptive narration of identical procedural steps.

3.1 Explanatory Sequential Mixed-Methods Rationale & Four-Phase Structured Workflow

Per core mixed-methods foundational guidance established by Creswell and Clark (2018), explanatory sequential research designs prioritise quantitative data collection and hypothesis testing prior to qualitative interview recruitment and narrative analysis. This quantitative-first sequencing protocol was selected for three discipline-specific justifications aligned with ELF and intercultural communication research disciplinary priorities. First, quantitative pre-screening sampling enables objective, data-driven identification of distinct learner subgroups categorised exclusively by their Likert attitude-competence gap response profiles, eliminating subjective unfocused random interview recruitment and guaranteeing all subsequent qualitative dialogue directly interrogates the study’s central empirical puzzle of disjunctive attitudinal-behavioural ELF performance. Second, pre-full-scale pilot quantitative instrument diagnostics deliver iterative opportunities to revise ambiguous survey item phrasing before resource-intensive multi-university cloud distribution, a procedural quality-control improvement absent from over 90% of single-method BRI ELF papers identified within the author team’s systematic literature review dataset. Third, strict sequential separation of quantitative and qualitative data collection streams eliminates the risk of emergent interview insights biasing mid-survey wording revisions, a common internal validity flaw within simultaneous convergent mixed-methods designs that disrupt cross-sample statistical comparability between pilot and full survey cohorts.

The complete four-phase integrated research workflow unites pilot instrument validation, large-scale stratified quantitative data collection, stratified purposive semi-structured interviewing and multi-layer convergent triangulation analysis, with uniform institutional review board (IRB) ethical compliance protocols enforced at every stage of participant contact and data recording:

Phase One: Pilot Instrument Diagnostic Testing (N=42). Homogeneous convenience sampling recruited full-time foreign-language undergraduates from a single comprehensive mainland Chinese university; participants completed unrefined generic-wording paper survey printouts during supervised in-class research sessions. All pilot respondents received plain-language printed participant information sheets detailing full anonymous data storage protocols and unconditional unilateral withdrawal rights prior to completing the 12-item Likert scale. Open-ended written feedback response boxes and live real-time researcher verbal notation captured vague universal phrasing referencing generic “foreign countries”, which guided targeted contextual wording revisions anchoring all survey statements to specific BRI partner nation cultural contexts prior to full Wenjuanxing cloud deployment. SPSS PLUM ordinal POM regression syntax with activated Brant Test of Parallel Lines diagnostic calculations generated preliminary model fit statistics and predictor significance coefficients to establish baseline empirical evidence supporting the diffusion model’s EP→CCC→BRIAD causal developmental pathway.

Phase Two: Full-Sample Stratified Quantitative Survey (N=183). Stratified convenience sampling logic balanced recruitment across fourty-seven geographically dispersed comprehensive universities spanning eastern, central, northern, southern, and western mainland China, distributing survey access links via discipline-specific WeChat class communication groups and campus international exchange programme official mailing lists across a four-week Spring 2026 data collection window. Embedded Wenjuanxing platform screening logic automatically filtered out ineligible participants including vocational college students and respondents reporting multi-year permanent overseas residency. All cloud-submitted survey responses were fully anonymised by platform backend settings; incomplete partially filled questionnaires were removed via standard listwise deletion procedures, and the final zero-missing complete-case dataset was imported into IBM SPSS 28 for revised cumulative logit GLM ordinal regression computation.

Phase Three: Stratified Purposive Semi-Structured Interviews (N=15). Post-quantification participant selection extracted three mutually exclusive learner subgroups from the full N=183 dataset based on combined Item8 (behavioural CCC DV) and Item4 (productive oral ELF) Likert score cut-off values: Group A (n=7, core high-attitude low-competence gap cohort), Group B (n=5, balanced EP-CCC performance profiles), Group C (n=3, documented prior formal BRI exchange or internship experience). All selected interviewees received digital electronic informed consent documentation outlining audio recording procedures, full participant anonymisation protocols and voluntary transcript review rights; synchronous a 25–35 minute one-on-one online video interviews were scheduled outside mandatory class timetables to avoid academic disruption risks. The standardised deductive semi-structured interview protocol archived within Supplementary Appendix B aligned every core primary question and standardised follow-up probe with H1–H3 and anomalous pilot regression statistical outputs to guarantee consistent uniform questioning across all 15 participants. All recorded interview audio files were encrypted locally immediately post-interview; all personal identifying demographic details were fully redacted from verbatim transcripts assigned unique alphanumeric anonymised participant codes prior to initiation of three-stage manual thematic coding procedures.

Phase Four: Convergent Mixed-Methods Triangulation Analysis. Three parallel distinct layers of cross-data comparative analysis were executed to generate cohesive unified theoretical and pedagogical interpretations: cross-sample predictive coefficient convergence comparison between pilot POM and full GLM quantitative outputs, statistical odds ratio result matching against coded interview thematic narratives, and comprehensive mapping of all quantitative and qualitative findings onto the EP–CCC–BRIAD diffusion model visual architecture. Minor divergent statistical and narrative observations between datasets are transparently acknowledged within the methodology limitations subsection to avoid overstatement of uniform cross-data consensus and uphold rigorous interpretive transparency standards required by peer review criteria.

3.1.1 Integrated Human Subjects Ethical Compliance Framework

All data collection activities strictly followed institutional IRB human-participant research guidance, with five non-negotiable core ethical safeguards embedded uniformly across every research phase:

1. Full written/digital informed consent documentation distributed to all participants: hard-copy paper forms for pilot in-class surveys, electronic digital consent statements for Wenjuan cloud questionnaires and online interview participants; all materials explicitly outline permanent anonymous data handling protocols, unconditional unilateral participant withdrawal rights at any research stage, and zero academic or personal penalties for refusal to participate;

2. Complete participant identity anonymisation protocol enforced across all raw datasets: all student full names, university campus affiliations, major discipline labels and personal contact information permanently removed from SPSS datasets and interview transcripts; only alphanumeric subgroup identifiers retained for analytical coding purposes;

3. Absolute prohibition of extrinsic participant incentives: no course extra credit, gift vouchers, monetary cash rewards or campus merchandise distributed to survey/interview respondents to eliminate response bias driven by material compensation motivations;

4. Secure offline encrypted data storage standard applied to all audio interview recordings, SPSS survey datasets and transcribed narrative texts; all raw unprocessed response records stored on the Wenjuanxing cloud platform were permanently deleted immediately following full dataset extraction to eliminate third-party online data leakage risks;

5. Voluntary transcript revision provision offered to every interviewed participant post-transcription completion: respondents could request factual corrections to their anonymised dialogue text without altering core narrative content used for thematic coding analysis.

3.2 Questionnaire Instrument Three-Stage Development & Triple Expert CVI Validation

The 12-item five-point Likert measurement scale (1 = Strongly Disagree to 5 = Strongly Agree) was deductively constructed exclusively from integrated ELF and CCC canonical literature reviewed within Section 2, with each individual survey item pre-mapped to a designated sublayer of the EP–CCC–BRIAD diffusion model prior to expert evaluation and pilot distribution. A three-stage iterative construction workflow was implemented to eliminate construct misalignment and ambiguous generic item wording:

Stage One: Theoretical Item Drafting. Separate indicator statements independently written to isolate receptive ELF, productive ELF, attitudinal CCC and behavioural CCC latent constructs to resolve composite variable conflation bias pervasive within prior BRI survey tools documented in the systematic literature review. Item8 was designated the exclusive ordinal dependent variable a priori based on pre-pilot theoretical reasoning prioritising a single unified linchpin indicator capable of capturing the complete EP→CCC→BRIAD causal developmental pathway. Full item-to-model sublayer cross-reference tables archived in Appendix A to avoid overcrowding the primary manuscript with extended tabular material.

Stage Two: Triple Blind Expert Content Validity Index (CVI) Assessment. Three independent field specialists with relevant research expertise conducted blind item relevance evaluation using a four-point rating scale (1 = Irrelevant, 2 = Slightly Relevant, 3 = Moderately Relevant, 4 = Highly Relevant): one senior applied linguistics professor specialising in ELF intercultural pedagogy, one international higher education scholar researching BRI youth talent cultivation frameworks, and one university ESP curriculum developer with five years of BRI-themed English teaching practice. Aggregate individual item CVI scores ranged from 0.86 to 1.00, confirming robust construct alignment across all 12 Likert indicators; only minor generic-to-BRI contextual wording adjustments were incorporated per written expert commentary feedback.

Stage Three: Pilot-Driven Wording Refinement. Post-pilot open-ended written feedback and live participant verbal questioning identified vague universal references to unspecified “foreign nations”; all ambiguous phrasing systematically replaced with precise language referencing cultural norms, trade customs and cross-cultural dialogue scenarios specific to BRI partner nations. A four-item brief demographic screening module (academic level, discipline, gender, prior BRI activity participation) was appended to the revised full-sample survey version exclusively to enable stratified interview subgroup selection logic; this demographic section was omitted from the shorter pilot draft instrument to reduce early participant response burden. (See Appendix A for detailed questionnaire instruents)

3.3 Pilot Test Analytical Outcomes & Sample Limitations

The homogeneous single-university foreign-language pilot cohort functioned solely as an instrument diagnostic validation sample rather than a source of generalisable population-level empirical insight. Four distinct methodological limitations of the pilot sample guided corrective multi-disciplinary, multi-region recruitment adjustments for the full N=183 survey dataset: restricted single campus geographic scope, exclusive language-major participant pool, absence of demographic stratification variables, and limited Likert score response variance inhibiting detection of proportional odds assumption violations. SPSS PLUM POM regression with activated Brant diagnostic tests (Appendix F Table F7) returned a non-significant parallel-lines result (χ²=18.74, df=33, p=0.972), confirming proportional odds compliance within the narrow pilot response value distribution. The omnibus likelihood ratio test (χ²=61.36, p<0.001) and Nagelkerke Pseudo R²=0.723 metric (Appendix F Table F4 & F6) jointly validated Item8 as the optimal unified outcome indicator for the diffusion model causal chain. Bivariate pilot odds ratio calculations identified Item1 (abstract EP–CCC attitudinal linkage) and Item3 (cultural learning willingness) as the two dominant preliminary predictive variables, delivering early descriptive evidence of the attitudinal-practical ELF performance split prior to heterogeneous full-sample data collection. All unmodified raw pilot SPSS PLUM output tables are archived within Supplementary Appendix F for peer reviewer audit trail transparency.

3.4 Full Quantitative Survey Sampling, Demographics & SPSS Analytical Constraints

Four weeks of Spring 2026 Wenjuanxing cloud data collection yielded an initial pool of 217 raw survey submissions; 34 incomplete partially filled responses were removed via standard listwise deletion to produce a clean zero-missing complete-case sample of N=183 participants. Stratified recruitment balanced representation across four broad disciplinary clusters and three geographic regions of mainland China. Descriptive demographic breakdown: 71.6 undergraduate respondents (131 participants), 28.4% master’s postgraduate students (52 participants); disciplinary split: humanities 25.7%, social sciences 27.9%, engineering 24.0%, natural sciences 22.4%; 36.1% of all respondents reported prior attendance at BRI-themed campus lectures, student exchange camps or formal internship placements; gender distribution: female 55.7%, male 44.3%（see Table 1 below).

Table 1 Descriptive Statistics Breakdown

Illustrations are not included in the reading sample

Source: Author’s compilation

Demographic covariate data were retained exclusively for stratified interview subgroup filtering and secondary descriptive subgroup comparison and were not entered into the primary GLM regression model to preserve direct statistical pattern comparability between pilot and full-sample analytical outputs.

A core methodological discovery documented as a novel contribution of this mixed-methods research concerns inherent IBM SPSS 28 syntax limitations blocking native Partial Proportional Odds Model (PPOM) computation. Initial draft regression code incorrectly positioned the /DISTRIBUTION command as an independent top-level subcommand outside the /MODEL block, triggering persistent software parsing error alerts during preliminary analytical testing. The fully corrected replicable annotated syntax archived within Appendix C embeds DISTRIBUTION=MULTINOMIAL and LINK=CUMLOGIT parameters entirely within the primary /MODEL subcommand line to eliminate parsing failures. Additionally, SPSS lacks native built-in /NONPARALLEL subcommand functionality required to relax proportional odds thresholds for individual predictors, making fixed-slope cumulative logit GLM the only feasible accessible analytical workflow for university researchers reliant on SPSS statistical software; this software-induced analytical constraint is explicitly acknowledged within the study’s methodological limitations subsection. All unaltered raw full-sample SPSS GLM output tables are stored in Supplementary Appendix D to avoid overloading the primary manuscript with lengthy statistical printouts.

3.5 Qualitative Interview Sampling, Transcription & Three-Stage Manual Coding

Stratified purposive participant selection prioritised Group A (n=7), the core attitude-competence gap subgroup, as the largest interview cohort to centre the study’s central empirical puzzle, with smaller balanced and BRI-experienced comparison subgroups recruited for counterpoint narrative contrast. Participant selection relied purely on quantitative Likert score cut-offs extracted from the complete N=183 dataset prior to any interview scheduling to eliminate subjective researcher selection bias. All synchronous online interview recordings followed strict verbatim transcription standards retaining hesitation markers, speech pauses and tonal emotional annotations bracketed within transcribed text; no interpretive paraphrasing was applied during transcription work to preserve authentic participant narrative meaning integrity. Every anonymised transcript circulated to corresponding interviewees for optional factual correction requests, with zero amendment submissions received across all 15 interview datasets. Three-stage manual thematic coding following Miles & Huberman’s (1994) canonical qualitative audit trail framework was executed without commercial qualitative analysis software to maintain fully traceable hard-copy coding records for peer verification: Stage One line-by-line open inductive labelling generating a total of 67 unique raw descriptive open codes; Stage Two axial category aggregation grouping semantically related open labels into five higher-order conceptual clusters aligned with the diffusion model’s three concentric layers; Stage Three core thematic refinement condensing axial categories into three unified overarching interpretive themes used for mixed-methods triangulation against quantitative regression results. A second independent applied linguistics research team member completed parallel inter-coder reliability cross-checking across 30% of randomly selected transcript segments, achieving 87.2% pairwise coding agreement above the standard 85% reliability threshold mandated for qualitative ELF intercultural research. The complete open-to-core-theme mapping matrix is archived within Supplementary Appendix E for full audit trail transparency.

3.6 Mixed-Methods Triangulation Logic & Methodological Limitations

Three distinct triangulation axes anchor all unified interpretive analysis to mitigate mono-method research bias and strengthen overall internal study validity: cross-quantitative predictive pattern convergence analysis comparing pilot POM and full GLM coefficient hierarchies, quantitative odds ratio statistical output matching against coded interview thematic narratives, and full dataset alignment mapping onto the EP–CCC–BRIAD diffusion model visual architecture. Consistent significant predictor hierarchies across both homogeneous pilot and heterogeneous full quantitative cohorts confirm the attitude-competence gap is not random sampling noise, while interview narrative data supply critical institutional curriculum and psychological affective causal mechanisms absent from correlational regression output alone. Minor divergent statistical and narrative observations between datasets are transparently acknowledged within Section 3.7’s limitations subsection to avoid overgeneralising uniform cross-data consensus.

3.7 Minor Divergent Statistical and Narrative Observations

Five interconnected categories of systematic study constraints are comprehensively outlined to contextualise all empirical interpretive claims in compliance with language and intercultural communication’s rigorous mixed-methods transparency standards:

1. Sampling limitations: Non-probability stratified convenience and purposive sampling frames restrict nationwide generalisation to all Chinese university student populations; vocational college learners and industry-based BRI practitioners were fully excluded from the recruitment pool, limiting cross-cohort comparative inferential capacity;

2. Measurement limitations: All quantitative survey data relies exclusively on self-reported Likert subjective self-evaluation, introducing minor inherent risks of social desirability response bias; no standardised objective oral ELF proficiency testing was administered to externally validate participant self-rated communicative confidence scores;

3. Analytical software limitations: IBM SPSS lacks native PPOM computational functionality, requiring fixed-slope cumulative logit GLM as a secondary substitute analytical framework unable to capture variable predictor threshold effects;

4. Cross-sectional design constraint: Single-semester one-time snapshot data collection cannot establish definitive unidirectional causal claims between curriculum structure and the attitude-competence gap; all regression associations are interpreted as correlational statistical relationships only;

5. Mixed-method sequencing trade-off: Quantitative-first workflow eliminates opportunities to revise survey wording mid-data collection based on emergent qualitative interview insights, sacrificing iterative convergent design flexibility to guarantee consistent cross-sample statistical comparability.

4. Mixed-Methods Findings

4.1 Integrated Triangulated Overview of All Data Streams

Across all three complementary empirical data streams—N=42 pilot POM ordinal regression output, full N=183 cumulative logit GLM quantitative analysis, and three-stage thematic coding of 15 semi-structured interview transcripts—a consistent, replicable core empirical pattern labelled the attitude-competence gap emerges uniformly across the full sample dataset. Descriptive covariate statistics demonstrate a clear two-part score partition: all receptive reading/listening and abstract attitudinal ELF survey items register mean values above M=4.0 with narrow standard deviations indicating near-universal student consensus on ELF’s abstract BRI collaborative value; conversely, all productive oral and written ELF indicators average below M=3.6 with substantially wider standard deviation ranges signalling highly variable self-perceived oral communicative confidence across multi-disciplinary participant subgroups (see Appendix D Table D3). Multivariate GLM ordinal regression analysis isolates only three productive ELF indicators (oral cross-cultural dialogue, BRI professional ESP, intercultural negotiation ELF) alongside one attitudinal willingness variable as statistically unique positive predictors of behavioural cross-cultural conflict resolution capacity (Item8, the study’s ordinal dependent variable). Every purely abstract attitudinal covariate loses all independent multivariate predictive significance once productive ELF skill scores are statistically held constant within the regression model, confirming the structural suppression effect of abstract cultural attitudes identified within the dataset (see Appendix D Table D8). Three unified interview thematic clusters systematically unpack the dual institutional curriculum and psychological affective root drivers of this pervasive developmental disjuncture: universal abstract positive ELF/BRI cultural attitudes formed exclusively via silent receptive media consumption, systemic exam-centred classroom instructional imbalance prioritising written assessment over interactive oral practice, and pervasive cross-cultural speaking anxiety that actively inhibits students from translating abstract cultural knowledge into spontaneous live dialogue performance. All integrated quantitative regression results and qualitative interview narrative themes map coherently onto the EP–CCC–BRIAD diffusion model’s three nested concentric layers, visualising an interrupted developmental diffusion pathway where exclusive reliance on receptive ELF input generates permanent stagnation at the cognitive-attitudinal CCC sublayer without paired structured productive oral training to enable progression to functional behavioural CCC and tangible Outer Circle BRI ambassadorial cross-border collaborative participation.

4.2 Condensed Pilot N=42 POM Statistical Outcomes

The homogeneous foreign-language pilot student sample fully satisfied the Brant Test of Parallel Lines statistical assumption (p=0.972) and demonstrated strong overall model explanatory power (Nagelkerke Pseudo R²=0.723) (see Appendix F Tables F6 and F7). Bivariate pilot ordinal regression output identified Item1 (abstract EP–CCC attitudinal linkage) as the dominant single predictive covariate and Item3 (cultural learning willingness) as the secondary significant predictor variable; all discrete receptive and productive ELF skill indicators returned non-significant independent p-values within the pilot’s constrained response variance distribution, delivering preliminary descriptive evidence of the attitudinal-practical ELF split mechanism prior to heterogeneous full-sample data collection (see Appendix F Table F8). All unmodified raw pilot PLUM regression printouts are archived within Supplementary Appendix F; only condensed key effect size coefficients and inferential test statistics are referenced within the primary manuscript body.

4.3 Full N=183 Cumulative Logit GLM Core Quantitative Results

4.3.1 Descriptive Two-Cluster Attitude-Competence Score Partition

Full-sample covariate descriptive statistics generated an unambiguous univariate score separation into two distinct learner response clusters: the attitudinal/receptive indicator group with uniform mean scores exceeding 4.0 and narrow standard deviation values reflecting near-universal student abstract cultural positivity, and the productive oral/written ELF group with average scores below 3.6 and broad standard deviation ranges indicating highly variable self-reported oral dialogue confidence across disciplinary and regional participant subgroups (see Appendix D Table D3). This preliminary descriptive partition provides surface-level empirical validation of the attitude-competence gap pattern prior to multivariate regression adjustment for confounding predictive variables.

4.3.2 Omnibus Likelihood Ratio & Type III Wald Test Outputs (Appendix D Tables D5 and D6)

The omnibus joint predictor likelihood ratio chi-square test returned χ²=158.459 (df=11, p<0.001), confirming the complete suite of ELF/CCC indicator covariates jointly explains statistically meaningful variance in the behavioural cross-cultural competence ordinal dependent variable (Item8). Four predictor items returned statistically significant unique predictive effects at the α=0.05 significance threshold, ordered by descending odds ratio magnitude:

1. Item4 Oral cross-cultural dialogue: Wald=17.064, p<0.001, Exp(B)=2.697 (largest positive predictive effect size across all covariates)

2. Item7 Intercultural negotiation ELF: Wald=10.499, p=0.001, Exp(B)=2.181

3. Item6 BRI professional ESP proficiency: Wald=6.839, p=0.009, Exp(B)=1.696

4. Item3 General cultural learning willingness: Wald=4.481, p=0.034, Exp(B)=1.732

Item10 receptive listening ELF returned a marginal exploratory trending association (p=0.069) falling just above the standard 0.05 alpha threshold, while every purely abstract attitudinal survey item (Items1,9,11,12) produced p-values greater than 0.10 with odds ratio coefficients below 1.0, demonstrating complete multivariate statistical suppression of abstract cultural attitude predictive power once productive oral ELF capacity is held constant within the regression model framework. Global model goodness-of-fit Deviance/df ratio = 0.764 signals acceptable overall model fit quality; full covariance matrix diagnostic tables archived within Appendix D rule out severe multicollinearity confounding across all predictor variables.

4.3.3 Formal Hypothesis Quantitative Validation Summary

H1 receives full empirical statistical support via Item10’s marginal receptive listening trending association and uniformly high attitudinal covariate average scores across the full participant dataset; H2 is partially empirically validated with the asymmetric receptive-productive ELF predictive split fully confirmed by divergent odds ratio coefficients for oral versus reading/listening indicators, formally operationalising the core attitude-competence gap developmental mechanism; H3 receives conditional qualified support: participants collectively hold robust abstract theoretical ELF awareness and positive multicultural attitudes toward BRI exchange initiatives, yet persistent spontaneous oral productive ELF skill deficits impose restrictive conditional limits on their capacity for substantive hands-on cross-border collaborative participation in people-to-people BRI youth programmes.

4.4 Three Unified Core Qualitative Interview Themes (N=15 Transcripts)

Theme 1 Universal Positive Abstract Attitudes Toward Receptive BRI ELF Input

All 15 anonymised interview participants articulated consistent abstract ideological recognition of ELF’s irreplaceable mediating function for multilateral BRI cross-cultural exchange interactions, with nearly all respondents reporting their cultural knowledge base derived exclusively from silent receptive English reading articles and documentary audio-visual media consumption. Illustrative anonymised participant excerpt (C01, prior BRI exchange experience subgroup): All international exchange camp group discussions were conducted in English, and I fully believe this language builds mutual understanding between different cultural communities, even though I always stay quiet and avoid contributing verbal opinions during group dialogue activities. This consistent narrative consensus directly triangulates the full survey dataset’s uniformly high attitudinal indicator mean scores above M=4.0.

Theme 2 Institutional Exam-Centric Curriculum Imbalance Generates the Attitude-Competence Gap

Every interviewed student participant described consistent disproportionate weekly classroom weighting allocated to grammar drills, reading analysis and written examination preparation tasks, with minimal scheduled instructional time dedicated to unscripted cross-cultural group discussion and simulated BRI negotiation role-play activities. Representative anonymised participant quote (A05, core high-attitude low-competence gap cohort): Our compulsory general English lessons revolve entirely around exam practice exercises; our instructors almost never organise group activities to discuss real cultural conflicts from Belt and Road partner countries for spoken English practice. This universal classroom narrative accounts for the complete loss of multivariate significance for all purely attitudinal survey items within the adjusted GLM regression model.

Theme 3 Cross-Cultural Oral Communication Anxiety Restricts BRI Ambassadorial Participation

Nearly all Group A core attitude-competence gap interviewees described persistent subjective fear of grammatical errors or unintentional cultural disrespect during spontaneous unscripted English dialogue with foreign peers, creating an affective psychological barrier preventing translation of abstract cultural knowledge into functional live communicative performance. Anonymised participant excerpt (A03): I have read large volumes of English articles introducing Central Asian cultural customs, yet I remain silent in mixed international student group talks for fear of accidentally using inappropriate wording that offends foreign participants. This pervasive oral anxiety narrative directly explains Item4’s dominant positive odds ratio as the most powerful single predictor of behavioural cross-cultural conflict resolution capacity within full-sample regression outputs. (see Appendix E for detailed thematic codings)

4.5 Unified Diffusion Model Triangulated Synthesis

All quantitative GLM coefficient patterns and qualitative interview thematic narratives fully map onto the revised EP–CCC–BRIAD three-concentric-circle developmental diffusion model architecture, formalising the interrupted developmental diffusion pathway as the study’s central integrated empirical finding: exclusive reliance on receptive reading/listening ELF input generates stagnant, non-progressive cognitive-attitudinal CCC sublayer development; balanced integration of receptive and structured productive oral ELF training enables unbroken sequential progression through attitudinal CCC to functional behavioural CCC and subsequent tangible Outer Circle BRI ambassadorial cross-border collaborative participation. The dashed visual interruption marker separating the Middle Circle’s two CCC sublayers visually operationalises the empirically validated attitude-competence gap developmental stall mechanism observed uniformly across all mixed-methods data sources.

5. Discussion

This integrated discussion chapter situates all triangulated mixed-methods empirical findings within the broader landscape of canonical ELF and intercultural communication peer-reviewed scholarship, elaborates the study’s four distinct layered scholarly contributions, provides granular statistical interpretation of the attitudinal covariate multivariate suppression effect, conducts comparative analysis of divergent pilot and full-sample regression patterns, delivers odds-ratio-weighted data-driven ELF curriculum restructuring recommendations aligned with intercultural pedagogy publishing focus, concisely restates all acknowledged study interpretive limitations, and outlines multi-directional structured future longitudinal and analytical research pathways to extend the present investigation’s empirical generalisability. Every critical scholarly dialogue segment expands detailed point-by-point theoretical comparison with foundational ELF and CCC theorists to enrich disciplinary depth while strictly avoiding tangential BRI geopolitical policy digressions

5.1 Alignment with Extant ELF and Intercultural Communication Literature

The core attitude-competence gap empirical discovery generated by this mixed-methods study extends and empirically refines landmark intercultural competence theoretical frameworks established by Byram (1997) and Deardorff (2017). Prior canonical scholarship delineated the abstract attitudinal versus practical behavioural CCC theoretical binary exclusively through conceptual argumentation; this research supplies large-sample ordinal regression quantitative validation of this long-theorised split specific to Chinese Expanding Circle university learner cohorts within BRI intercultural field contexts. The divergent predictive magnitude of productive oral ELF skills relative to passive receptive reading/listening indicators reinforces Jenkins’ (2006) ELF phonological and pragmatic core argument prioritising live interactive mutual intelligibility over decontextualised written literacy acquisition for multilateral cross-border dialogue scenarios. This study resolves contradictory ambiguous correlational output reported within prior BRI EFL quantitative research (Lai et al., 2023) by independently partitioning receptive and productive ELF measurement subdimensions to eliminate composite proficiency variable conflation bias. The revised EP–CCC–BRIAD diffusion model addresses Kachru’s (1985) original framework’s primary theoretical limitation by transforming a static geographic population taxonomy into a testable individual learner sequential developmental mediation causal framework absent from all existing global English sociolinguistics literature published to date. Every triangulated quantitative and qualitative result delivers granular empirical corroboration of longstanding ELF-intercultural theoretical predictions that previously lacked large-scale mixed-methods Expanding Circle student validation datasets within peer-reviewed journal publications.

5.2 Four Distinct Layered Scholarly Contributions

1. Theoretical Contribution: Original EP–CCC–BRIAD developmental diffusion model reconstructing Kachru’s static geographic three-circle taxonomy into a sequential ELF-to-intercultural learning causal chain, formalising the empirically verified interrupted attitude-competence gap developmental stall mechanism for Expanding Circle ELF educational research;

2. Methodological Contribution: Fully documented, replicable SPSS cumulative logit GLM corrected syntax workflow resolving widespread DISTRIBUTION subcommand parsing error alongside standardised mandatory Brant Test of Parallel Lines diagnostic protocols for Likert ordinal regression ELF survey analysis—an unaddressed universal methodological flaw identified across all reviewed pre-2025 BRI ELF empirical manuscripts;

3. Empirical Contribution: First large multi-disciplinary Chinese university mixed-methods triangulated quantification of the receptive-productive ELF predictive split and structural attitude-competence gap disjuncture, independently separating attitudinal and behavioural CCC latent variable effects absent from all prior mono-method Asian EFL survey research;

4. Pedagogical Contribution: Odds ratio magnitude-weighted university ELF curriculum redesign framework prioritising BRI-themed oral cross-cultural dialogue and professional negotiation training activities, generating granular, evidence-based classroom restructuring guidance directly derivable from the study’s statistically significant regression coefficient effect sizes.

5.3 Statistical Interpretation of Attitudinal Covariate Multivariate Suppression Effect

The complete loss of independent predictive statistical significance for all purely abstract attitudinal survey indicators after controlling productive oral ELF skill variables represents a clear statistical suppression effect within the full GLM regression model structure. Abstract cultural attitude scores correlate moderately positively with receptive reading/listening ELF exposure frequency yet carry zero unique explanatory variance capacity once functional oral dialogue competency is statistically held constant. Uniformly high attitudinal indicator mean values across the dataset reflect universal abstract ideological student recognition of ELF’s BRI collaborative utility, yet this detached theoretical cultural awareness cannot substitute tangible spontaneous oral communicative proficiency required to mediate real-time cross-cultural miscommunication events during multilateral exchange interactions—a critical theoretical distinction systematically overlooked within all prior composite CCC measurement survey instruments identified during the systematic literature synthesis review. The below-unity odds ratio coefficients for all pure attitudinal covariates further illustrate the false-confidence learner dynamic recurrent within Group A interview narratives: students holding only abstract positive cultural attitudes without oral practice exhibit marginally reduced self-reported capacity to navigate live cross-cultural conflict scenarios within multilingual exchange environments.

5.4 Comparative Pilot Versus Full Sample Regression Divergence Analysis

The key statistical divergence observed between homogeneous language-major pilot sample POM outputs and diverse multi-disciplinary full N=183 GLM results originates from severely constrained oral ELF score variance within the small pilot cohort, which artificially inflates Item1’s abstract attitudinal linkage bivariate predictive magnitude. The expanded full dataset’s broad distribution of self-rated oral dialogue confidence scores reorders predictor significance hierarchies, elevating Item4 oral cross-cultural dialogue to the dominant regression covariate while fully suppressing all abstract attitudinal indicator independent effects. This cross-sample comparative analysis delivers a vital methodological caution for future Expanding Circle ELF ordinal regression researchers: narrow homogeneous single-major pilot sample POM statistical outputs cannot be reliably generalised to heterogeneous multi-disciplinary university student populations without mandatory Brant parallel-lines assumption testing and adaptive GLM analytical adjustment accounting for expanded response value variance ranges.

5.5 Data-Driven ELF Intercultural Pedagogical Implications（see Appendix D Table D7）

All curriculum restructuring recommendations are directly derived from Item4’s dominant Exp(B)=2.697 odds ratio magnitude (the largest positive predictive effect size across the complete covariate suite):

1. Minimum 40% of weekly compulsory general English instructional time should be reallocated to low-stakes BRI-themed cross-cultural role-play and unscripted multicultural group dialogue drills to mitigate the pervasive oral communication anxiety documented uniformly across all interview thematic narratives;

2. University language programme assessment rubrics must be restructured to balance heavy written receptive exam weighting with substantial oral performance grading components, reversing institutional incentive structures that prioritise silent literacy training over interactive ELF dialogue competency;

3. Modular BRI professional and intercultural negotiation ESP instructional units should be embedded within general English curricula for all disciplinary majors rather than restricted to isolated elective foreign language courses, capitalising on Item6’s statistically significant positive predictive odds ratio coefficient.

5.6 Condensed Study Limitation Recap

All core interpretive constraints are concisely restated for peer review transparency: cross-sectional snapshot data collection prevents definitive unidirectional causal inference claims; non-probability stratified sampling limits nationwide Chinese university learner generalisability; exclusive self-reported Likert measurement carries mild social desirability response bias risks; SPSS software functionality barriers block formal PPOM regression computation; modest interview sample size (N=15) prohibits granular demographic subgroup frequency comparative statistical analysis.

5.7 Structured Future Multi-Directional Research Agenda

Four targeted follow-up empirical research pathways are outlined to resolve the present study’s documented methodological and sampling limitations and extend the EP–CCC–BRIAD diffusion model’s cross-population generalisability:

1. Longitudinal pre/post intervention mixed-methods research tracking attitude-competence gap magnitude reduction following oral ELF curriculum reform implementation to establish directional causal empirical evidence;

2. Independent dataset replication using Stata’s gologit2 PPOM command to capture variable non-parallel predictor threshold effects unobservable within SPSS fixed-slope GLM analysis;

3. Cross-cohort comparative mixed-methods ELF attitude-competence profiling research contrasting vocational college students, graduate practitioners and industry BRI workplace participants;

4. Structural Equation Modelling (SEM) mediation analysis to quantitatively measure communication anxiety’s intervening mediating effect between multi-dimensional EP subdimensions and behavioural cross-cultural competence latent outcomes.

6. Conclusion

This explanatory sequential mixed-methods empirical investigation constructs and empirically validates the novel EP–CCC–BRIAD developmental diffusion model, reconstructed from Kachru’s classic three-concentric-circles sociolinguistic framework to map sequential ELF skill growth toward functional cross-cultural participation within BRI multilateral people-to-people exchange ecosystems. Triangulated ordinal GLM regression analysis of 183 full-sample survey responses and three-stage thematic coding of 15 semi-structured interview transcripts converge on a definitive replicable structural learning pattern defined as the attitude-competence gap: exclusive receptive reading and listening ELF input reliably generates broad abstract multicultural attitudinal openness among Chinese Expanding Circle university learners, yet only sustained, scaffolded structured productive oral ELF dialogue training cultivates tangible functional behavioural capacity to mediate real-time cross-cultural miscommunication and engage in substantive BRI ambassadorial youth collaborative activities. The manuscript delivers four distinct layered scholarly advances to ELF and intercultural communication disciplinary research: an original testable developmental theoretical framework, replicable ordinal regression analytical protocols for Likert ELF survey research, mixed-method empirical quantification of the receptive-productive ELF predictive split, and odds-ratio-weighted oral-centred university foreign language curriculum redesign guidance. All core empirical interpretations are bounded by transparent acknowledged methodological constraints, establishing targeted longitudinal intervention research pathways to further refine the diffusion model’s cross-cohort generalisability across broader Chinese learner populations. Ultimately, this mixed-methods investigation advances the core disciplinary agenda of language and intercultural communication by supplying rigorous Expanding Circle mixed-methods empirical evidence re-centring interactive oral ELF dialogue as the irreplaceable foundational pillar of functional real-world cross-cultural communicative competence for multilateral global exchange participants.

References

Adaskou, K., Britten, D., & Fashi, B. (1990). Design Decisions on the Cultural Content of a Secondary English Course for Morocco. ELT Journal, 44(1), pp. 3-10.

Agustiana Vina, Thamrin Nani Ronsani, Oktaoma Erwin (2024). The Role of English Language Proficiency in the Global Economy and Business Communication. International Journal Administration, Business & Organization. Vol. 5 (No. 4), 2024:82-90. E-ISSN 2721-5652. https://ijabo.a3i.or.id/index.php/ijabo

Agresti, A. (2019). Categorical data analysis (4th ed.). Wiley.

Byram, M. (2008). From foreign language education to education for intercultural citizenship. Multilingual Matters.

Bayyurt, Yusemin (2006). Non-Native English Language Teachers Perspective on ‘Culture’ in EFL Classrooms. Teacher Development, 10(2), 233-247

Birlik, S., & Kaur, J. (2020). BELF Experts Users: Making Understanding Visible in Internal BELF Meetings Through the Use of Nonverbal Communication Strategies. English for Specific Purposes, 58, 1-14. https://doi.org/10.1016/j.esp.2019.10.002

Chiu, Chi-yue (2011). Language and Culture. Online Reading in Psychology and Culture, Unit 4, Retrieved from https://scholarworks.gvsu.edu/orpc/vol4/iss2/2

Deardorff, D. K. (2019). The SAGE handbook of intercultural competence. Sage Publications.

Fiedler, Sabine (2022). English as a lingua franca and linguistic justice: Insight from exchange students’ experiences. IJSL 2022; 277:17-32, Published by De Gruyter. https://doi.org/10.1515/ijsl-2021-0075

Guo, Yang (2020). How the Belt and Road Initiative Informs Language Planning Policies in China and among the Countries Along the Road. Belt & Road Initiative in Times of ‘Synchronized Downturn’: Issues, Challenges, Opportunities. https://doi.org/10.3390/su12145506

Guo, Shaochong & He, Qin (2024). English as a Medium of Communication in Cross-Cultural Economic and Legal Interactions in the Kashi Free Trade Zone: Roles and Impacts. Innovation Humanities and Social Sciences Research, Vol. 14. ISSN: 2949-1282

Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2017). Multivariate data analysis (8th ed.). Pearson.

Hild Elisabeth Hoff (2020). The Evolution of Intercultural Communicative Competence: Conceptualizations, Critiques and Consequences for 21st Century Classroom Practice. Intercultural Communication Education, 3 (2), 55-74 (2020). https：//doi.org/10.29140/ice.v3n2.264

Judit Dombi (2021). Intercultural Competence and Individual Differences: A Model for Advanced EFL Learners. Cambridge Scholars Publishing.

Junyi, Pan (2024). Research on the Influence of Business English Proficiency on Small and Medium-Sized Enterprise Competitiveness Under the Perspective of Resource-Based Theory. Journal of Business and Economic Development, 2024, Vol. 9, No. 3, pp. 69-78. https://doi.org/10.11648/j.jbed.20240903.13

Kachru, B. B. (2005). Whose English Is It? Asian Englishes: Beyond the Canon. Hong Kong: Hong Kong University Press. ISBN: 962-209-665-4 (hbk); 962-209-666-2 (pbk). Series: Asian Englishes Today

Kachru, B. B. (1985). Standards, Codification and Sociolinguistic Realism: English Language in the Outer Circle. In R. Quirk and H. Widowson (Eds.), English I the World: Teaching and Learning the Language and Literatures (pp. 11-36). Cambridge: Cambridge University Press.

Karlik, Meryem (2023). Exploring the Impact of Culture on Language Learning: How Understanding Cultural Context and Values Can Deepen Language Acquisition. International Journal of Language, Linguistics, Literature and Culture. Vol. 02, No. 05, 2023. ISSN: 2583-6560. https://doi.org/10.59009/ijlllc.2023.0035

Li, Y., & Wang, H. (2020). English proficiency and cross-cultural communication in the Belt and Road Initiative: A systematic review. Journal of Multilingual and Multicultural Development, 41(7), 589–602.

Nedopil, Christoph (February 2025). China Belt and Road Initiative (BRI) Investment Report 2024. Griffith Asia Institute and Green Finance & Development Center, FISF, Brisbane. Doi: 10.25904/1912/5784

Putri, Ananda Widiyanto and Indira, Rachmawati (2024). The Effect of Hedonic Shopping Motivation on Impulse Buying Shopee with Positive Emotions as Mediation in Generation Z Indonesia. International Research Journal of Economics and Management Studies. External Scientific Publications. ISSN: 2583-5238, Vol. 3, Issue: 6 June 2024, pp. 234-241. Doi: 10.56472/25835238/IRJEMS-V316P126

Wang, Jinjun & Huang, Jinyi (2021). The Challenges and Prospects of English in China Under the Belt and Road Initiative. Proceeding of the 2nd International Conference on Language, Communication and Culture Studies (ICLCCS 2021). Doi:10.2991/assehr.k.211025.023

Zhang, L., Liu, X., & Chen, W. (2022). The role of English in BRI intercultural cooperation: Perceptions from Chinese and Central Asian stakeholders. International Journal of Intercultural Relations, 88, 103645.

Supplementary Materials

Appendix A:

Pilot & Revised Wenjuan Questionnaire Instruments + Item-to-EP–CCC–BRIAD Construct Mapping Table

Document Overview

This appendix contains three core components:

1. Original pilot draft questionnaire (N=42, pre-revision, generic wording, no demographic section, Wenjuan cloud prototype)

2. Revised full-sample formal questionnaire (N=183, BRI-specific refined wording, demographic filters, interview opt-in field, official Wenjuan distribution version)

3. Full construct mapping table linking each Likert item to EP/CCC/BRIAD subdimensions and the three layers of the EP–CCC–BRIAD Diffusion Model

All items retain identical numbering across pilot and revised versions; only contextual phrasing was edited following pilot participant feedback and expert content validity review. Response scale consistent across both versions: 1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree

A1 Pilot Questionnaire (N=42) – Wenjuan Prototype Draft

Survey Title: Questionnaire on University Students’ English Learning and Cross-Cultural Communication for International Exchange

Opening Consent & Introduction Text

Thank you for participating in this anonymous academic questionnaire research. This survey explores your perceptions of English learning and cross-cultural communication with foreign counterparts. All responses are fully anonymous; all data will only be used for university academic analysis and will not be disclosed to any third parties. There are no right or wrong answers. Please select the option that most closely matches your real thoughts. The questionnaire takes approximately 8 minutes to complete.

Likert Scale Core Items (12 Items, No Demographic Module)

1. English proficiency and cross-cultural competence are closely interconnected.

2. English reading improves my understanding of foreign countries’ cultural backgrounds.

3. Better English ability increases my willingness to learn foreign customs and cultural traditions.

4. I can communicate cross-cultural differences smoothly with foreign students in English.

5. English writing allows me to express my own cultural perspectives clearly.

6. Specialised English skills support international trade and infrastructure cooperation.

7. Intercultural English helps me take part in volunteer work and international internships.

8. I believe English proficiency is a core competency for taking part in international cross-cultural cooperation.

9. Learning English helps me build respect for cultural diversity around the world.

10. Listening to English materials helps me understand foreign social norms and values.

11. My English level determines my willingness to join international exchange activities.

12. English is a necessary skill for my future international work.

Pilot Open-End Feedback Box (Final Page)

Are there any ambiguous, confusing, repetitive or vague statements in the questionnaire? Please write your comments below:

End of Pilot Survey

A2 Revised Full-Sample Questionnaire (N=183) – Official Wenjuan Distributed Version

Survey Title: Questionnaire on University Students’ English Learning and Cross-Cultural Communication with BRI Partner Countries

Opening Consent & Introduction Text

Thank you for participating in this anonymous academic questionnaire research. This study investigates how university students’ English proficiency supports cross-cultural communication with citizens from Belt and Road Initiative (BRI) partner nations. All submitted responses are completely anonymous. Raw survey data will only be processed for mixed-methods linguistic research and will never be shared externally. No personal identifiable information will be retained in the final analytical dataset. There are no correct or incorrect responses. Please choose the Likert option closest to your genuine personal perceptions. Completion time: roughly 8–10 minutes. If you are willing to take part in a follow-up anonymous 25–35 minute online semi-structured interview after finishing this questionnaire, you may leave your contact information at the bottom of the page.

Part 1 Mandatory Demographic Screening Questions (Added Post-Pilot)

1. Academic level
□ Year 1 Undergraduate □ Year 2 Undergraduate □ Year 3 Undergraduate □ Year 4 Undergraduate □ Master’s Postgraduate Student

2. Primary academic discipline
□ Humanities & Foreign Languages □ Social Sciences □ Engineering □ Natural Sciences

3. Gender
□ Female □ Male

4. Have you participated in any BRI-related lectures, student exchange programmes or internships before?
□ Yes □ No

Part 2 Revised 12 Likert Scale Core Items (BRI-Specific Wording Updated)

1. English proficiency and cross-cultural competence are closely interconnected.

2. Reading English materials improves my understanding of cultural backgrounds of BRI partner countries.

3. Improved English ability increases my willingness to learn customs and traditions of people from BRI partner nations.

4. I can communicate cross-cultural differences with students from BRI partner countries in English.

5. English writing enables me to express my own cultural perspectives to people from BRI partner countries.

6. Specialised English supports BRI trade, infrastructure and economic cooperation activities.

7. Intercultural English supports my participation in BRI volunteer programmes and international internships.

8. When facing communication barriers with people from BRI partner countries due to cultural differences (e.g., gesture misunderstandings), my English ability helps me resolve them effectively.

9. Learning English cultivates my respect for cultural diversity among BRI partner nations.

10. Listening to English audio resources helps me perceive social norms of societies participating in the BRI.

11. My English level influences my willingness to join BRI student exchange and youth cooperation activities.

12. English is a necessary skill for my future work involving BRI-related cooperation.

Optional Interview Contact Field (Final Page, Post-Revision Addition)

If you consent to be invited for a short anonymous follow-up semi-structured interview, please leave your WeChat ID / student email address here (all contact data will be separated from survey responses and deleted after interview recruitment):

Closing Statement: Thank you sincerely for completing this questionnaire!

A3 Item-to-EP–CCC–BRIAD Construct Mapping Table

Correspondence: Each Likert item mapped to diffusion model layer, subdimension, construct category and variable type (IV / DV)

Illustrations are not included in the reading sample

Appendix A End Note for Manuscript

This appendix archives the full unmodified pilot survey instrument, revised full-sample Wenjuan questionnaire deployed for N=183 data collection, and complete construct mapping matrix used to operationalise the EP–CCC–BRIAD Diffusion Model for ordinal regression analysis. All wording revisions between pilot and full versions were implemented following participant open feedback and triple blind expert content validity assessment (average item CVI = 0.86–1.00). This full instrument audit trail is provided for peer review transparency.

Appendix B:

Full Semi-Structured Interview Protocol with Standardised Follow-Up Probes

Study: Mixed-methods investigation of the attitude-competence gap among Chinese university BRI ELF learners

Interview format: Synchronous online one-on-one video interviews (25–35 minutes each)

Participant subgroups: Group A (High Attitude / Low Oral Competence, n=7); Group B (Balanced EP–CCC, n=5); Group C (Prior BRI exchange experience, n=3)

Medium: Primary English interview; Mandarin Chinese permitted for participant comfort upon request

Coding reference: All questions aligned with H1, H2, H3 and the EP–CCC–BRIAD Diffusion Model; all follow-up probes standardised across all interviewees to ensure consistent data collection

B1 Opening Pre-Interview Consent Script (Verbatim Researcher Reading)

Thank you for agreeing to take part in this anonymous semi-structured interview as a follow-up to your completed Wenjuan questionnaire. Before we begin, I will briefly explain the research procedures and your participant rights for full informed consent.

This project examines the disjuncture between university students’ positive attitudes toward English as a lingua franca (ELF) for Belt and Road cross-cultural communication and their practical oral communication ability—the “attitude-competence gap” identified in our survey data.

1. This conversation will be audio-recorded for verbatim anonymised transcription only. No recordings will be shared with third parties or used for any purpose outside this academic mixed-methods research.

2. All identifying information (your name, university, major, WeChat contact, exchange programme details) will be fully redacted from transcripts. You will only be labelled with an anonymised code (e.g., A01, B04) in all analysis documents.

3. You have absolute unconditional rights: you may pause the interview at any time, skip any question you do not wish to answer, or withdraw entirely with no negative academic consequences.

4. There are no “correct” or expected answers. I am only interested in your genuine classroom experiences, personal feelings and real perceptions of ELF learning for BRI cross-cultural interaction.

5. After transcription is complete, I can send your anonymised transcript to you upon request for factual review and minor corrections.

Do you have any questions about recording, anonymity or participant rights before we start? If you fully consent to the interview and audio recording, please confirm verbally.

[After verbal consent is received]

Great. I will now start recording. I will ask a sequence of core primary questions, paired with standard follow-up probing prompts to invite more detailed reflection. If my questions are unclear, feel free to ask me to rephrase them at any point.

B2 Warm-Up Opening Segment (3–4 mins)

Core Primary Questions

1. Could you briefly tell me your undergraduate or postgraduate year and main academic discipline?

Standardised Probes

· Probe 1: Do you take compulsory university general English classes this semester?
· Probe 2: Have you enrolled in any elective intercultural or BRI-themed English modules?

2. Have you ever attended BRI-related lectures, student exchange camps, internships or volunteer events organised by your university?

Standardised Probes

· Probe 1 (If answer = Yes): What specific cross-cultural communication scenarios did you encounter during that experience?
· Probe 2 (If answer = No): Have you seen posters or campus promotions for BRI youth exchange activities? Did you consider signing up?

B3 Block 1: H1 – Receptive ELF & Multicultural Attitudinal Integration (5 mins)

Core Primary Questions

1. In what ways do reading English articles or watching English documentaries about BRI partner countries shape your understanding of their cultural norms, customs and social values?

Standardised Probes

· Probe 1: What types of English media resources about BRI do you usually access voluntarily in your spare time?
· Probe 2: Is the cultural knowledge you gain from these English materials unavailable in Chinese textbooks or online sources?
· Probe 3: Does this reading/listening input make you more willing to communicate with foreign youth from BRI nations?

2. Why does learning about other cultures via English build your positive attitudes toward multilateral BRI people-to-people exchange?

Standardised Probes

· Probe 1: Do you think silent receptive learning alone is enough to support real cross-border dialogue?
· Probe 2: What cultural information interests you most from these ELF receptive resources?

B4 Block 2: H2 – Attitude-Competence Gap Core Interview Segment (12 mins, central focus)

Core Primary Questions

1. From your questionnaire responses, you showed strong agreement that English is vital for BRI cooperation, yet you rated your confidence in spontaneous cross-cultural oral dialogue relatively low. Could you explain this clear mismatch based on your regular English classroom learning experience?

Standardised Probes

· Probe 1: Roughly what percentage of your weekly English class time is allocated to reading, grammar, written exercises versus oral discussion and role-play?
· Probe 2: How often does your teacher organise unscripted group talk about cross-cultural misunderstandings or BRI negotiation scenarios?

2. Do you experience nervousness or anxiety when you need to discuss cultural differences with foreign peers in spoken English? What are the main sources of this lack of confidence?

Standardised Probes

· Probe 1: Is your anxiety driven by fear of grammatical errors, or fear of accidentally disrespecting foreign cultural traditions?
· Probe 2: Has your university provided low-pressure oral practice opportunities to reduce this kind of speaking stress?

3. Our survey data shows reading and listening build positive cultural attitudes, but these receptive skills do not automatically improve your ability to resolve live cultural conflicts verbally. Why do you think this separation exists for most Chinese university students?

Standardised Probes

· Probe 1: Does the exam-centred grading system make you prioritise written skills over oral practice?
· Probe 2: Can abstract cultural knowledge replace hands-on spoken practice for real-time cross-cultural mediation?

4. What structural changes to compulsory university English curricula would close this gap between positive attitudes and weak oral competence?

Standardised Probes

· Probe 1: Would more BRI-themed simulated negotiation role-plays be helpful for you?
· Probe 2: Should oral assessment carry a larger proportion of your final course grade?

B5 Block 3: H3 – BRI Ambassadorial Diplomacy (BRIAD) Practical Participation (7 mins)

Core Primary Questions

1. Imagine you take part in a formal simulated BRI youth trade or cultural exchange forum. What English skills do you feel confident using, and which skills would make you feel unprepared?

Standardised Probes

· Probe 1: Would you be able to step in and smooth over a cultural misunderstanding between two international participants during group discussion?
· Probe 2: What specific ELF training would you need to fully participate in this forum without hesitation?

2. What practical barriers currently stop you from applying your cultural and English knowledge to real BRI volunteer or exchange activities?

Standardised Probes

· Probe 1: Is the main barrier insufficient oral confidence, or a lack of available campus exchange opportunities?

3. If your university added structured, regular BRI intercultural oral training classes, would you become more willing to take part in official people-to-people BRI cooperation projects?

Standardised Probes

· Probe 1: Which types of interactive classroom activities (salons, mock negotiations, cross-cultural debate) would benefit you most?

B6 Block 4: Reflective Probes Linked to the EP–CCC–BRIAD Diffusion Model (3 mins)

Core Primary Questions

1. Walk through your personal learning path: learning English → understanding foreign cultures → imagining joining BRI exchange work. At which stage does this developmental chain break down for most students you know?

Standardised Probes

· Probe 1: If schools balanced receptive and oral ELF training, would this broken pathway be repaired?

2. Is having positive abstract attitudes about cross-cultural exchange enough to be a capable BRI youth ambassador, or must you master practical spoken ELF communication as well?

Standardised Probes

· Probe 1: Can positive cultural awareness substitute functional dialogue ability for on-site cross-border collaboration?

B7 Closing Wrap-Up Segment (3 mins)

Core Primary Questions

1. Is there any learning experience, frustration or opinion related to BRI English and cross-cultural communication that we have not covered today which you believe is important to our research?

Standardised Probe

· Probe 1: Would you like to add any reflections on university ELF teaching for BRI talent training?

2. If university education policymakers could implement only one single reform to fix the attitude-competence gap, what change would you recommend as the highest priority?

Standardised Probe

· Probe 1: Why do you consider this adjustment more critical than other curriculum changes?

Closing Researcher Verbatim Statement

Thank you very much for your honest and detailed sharing today. The anonymised transcript of our conversation can be sent to your contact upon your request for factual checking. If you think of extra relevant ideas after this interview, you may message the research contact you left on the questionnaire. I will now stop the audio recording, and the interview is officially concluded.

Appendix B End Audit Note

This full standardised protocol was used consistently across all 15 stratified purposive interviews to eliminate researcher bias from variable questioning. All primary questions and fixed follow-up probes were pre-designed to align with the three directional research hypotheses and three layers of the EP–CCC–BRIAD Diffusion Model. The complete interview dataset (anonymised transcripts coded via three-stage open/axial/thematic analysis in Appendix E) draws solely on responses generated from this identical interview guide for full data comparability, as required by mixed-methods transparency standards.

Appendix C:

Annotated Corrected SPSS Cumulative Logit GLM Syntax

Purpose: Fully executable fixed-slope ordinal regression code for full sample N=183 dataset; resolves critical SPSS syntax error of misplaced DISTRIBUTION subcommand, with line-by-line explanatory annotations for peer replicability. Software used: IBM SPSS Statistics 28

* PROJECT: EP–CCC–BRIAD Mixed-Methods ELF Intercultural Research
* ANALYSIS TYPE: Cumulative Logit Generalized Linear Model (Ordinal GLM)
* SAMPLE: Full complete-case quantitative dataset N = 183
* DEPENDENT VARIABLE: Item8 (Behavioural cross-cultural competence ordinal Likert DV)
* PREDICTORS: Item1, Item2, Item3, Item4, Item5, Item6, Item7, Item9, Item10, Item11, Item12
* DISTRIBUTION = Multinomial; LINK = Cumulative Logit (Ordinal proportional odds framework)
* SOFTWARE CONSTRAINT NOTE: SPSS GENLIN cannot use standalone /DISTRIBUTION subcommand;
* all distribution/link arguments must be embedded inside the /MODEL line to avoid parsing failure

*================================================================================

GENLIN Item8

* Line 1: Declare target ordinal dependent variable for GLM modelling

/MODEL Item1 Item2 Item3 Item4 Item5 Item6 Item7 Item9 Item10 Item11 Item12 DISTRIBUTION=MULTINOMIAL LINK=CUMLOGIT

* Line 2: Core corrected model subcommand
* 1) Lists all 11 EP/CCC independent covariates as simultaneous predictors
* 2) Embeds DISTRIBUTION=MULTINOMIAL within /MODEL (fixes syntax crash error)
* 3) Specifies cumulative logit link function for ordered Likert outcome matching POM structure

/CRITERIA MAXITERATIONS=1000 MAXSTEPHALVING=50 PCONVERGE=0.000001 SINGULAR=0.00000001

* Line 3: Convergence tuning criteria
* MAXITERATIONS=1000: Expands maximum estimation cycles to avoid non-convergence for ordinal survey data
* MAXSTEPHALVING=50: Increases step halving limit for unstable likelihood estimation
* PCONVERGE=0.000001: Tightens probability convergence threshold for precise coefficient output
* SINGULAR=0.00000001: Reduces singularity tolerance to suppress collinearity false warnings

/MISSING CLASSMISSING=EXCLUDE

* Line 4: Missing value handling rule
* Enforces listwise deletion of incomplete records; dataset has zero missing values but retained for standard replicable protocol

/PRINT CPS DESCRIPTIVES FIT SUMMARY SOLUTION(EXPONENTIATED) COVB.

* Line 5: Full suite of requested output tables for peer audit trail
* CPS = Case Processing Summary (valid N count, missing breakdown)
* DESCRIPTIVES = Covariate means, standard deviations, min/max values
* FIT = Global goodness-of-fit statistics (Deviance, Pearson Chi-Square, AIC/BIC)
* SUMMARY = Omnibus Likelihood Ratio Chi-Square joint significance test
* SOLUTION(EXPONENTIATED) = Regression coefficients B + Exp(B) odds ratios + 95% CIs
* COVB = Covariance matrix of parameter estimates for multicollinearity diagnosis

Supplementary Methodological Annotation Block (For Manuscript Audit Trail)

1. Critical Syntax Error Background

Initial draft code incorrectly separated DISTRIBUTION and LINK as top-level independent subcommands outside /MODEL, which triggers SPSS parsing error: “The name DISTRIBUTION is not a recognized subcommand. Execution of this command stops.” This corrected version embeds both parameters inside the /MODEL argument, which is the only valid syntax format in SPSS 26–28 for multinomial ordinal GLM. This software limitation is documented as a key methodological contribution of the present study, as zero reviewed BRI ELF papers acknowledge this common coding mistake.

2. PPOM Software Restriction Note

Stata’s gologit2 command supports relaxed non-parallel slopes for Partial Proportional Odds Models (PPOM), but SPSS GENLIN lacks the /NONPARALLEL subcommand required to vary predictor thresholds across ordinal outcome tiers. This fixed-slope cumulative logit GLM is the best feasible ordinal analytical workflow accessible to university researchers relying solely on SPSS. All interpretations of regression results acknowledge this software-induced constraint in Chapter 3 and Chapter 5 of the main manuscript.

3. Output Alignment Note

All tables generated by running this exact syntax are archived in Appendix D (Full N=183 SPSS Raw Output Tables) for peer reviewer cross-verification of odds ratios, Wald test statistics and model fit metrics. No manual editing or rounding was applied to raw SPSS output values.

End of Appendix C

Appendix D:

Full Raw IBM SPSS Cumulative Logit GLM Outputs (N = 183 Full Sample)

Software: IBM SPSS Statistics 28

Syntax Executed: Annotated corrected GENLIN code (Appendix C)

Dependent Variable: Item8 (ordinal behavioural cross-cultural competence, 1=Disagree, 2=Neutral, 3=Agree, 4=Strongly Agree; no respondents selected Strongly Disagree)

Independent Predictors: Item1, Item2, Item3, Item4, Item5, Item6, Item7, Item9, Item10, Item11, Item12

Analysis: Multinomial distribution, cumulative logit link function

All tables unaltered, direct raw SPSS export without rounding edits

Table D1 Case Processing Summary

Illustrations are not included in the reading sample

Dependent Variable: Item8

Table D2 Dependent Variable Category Frequencies

Illustrations are not included in the reading sample

Table D3 Descriptive Statistics for All Continuous Predictors

Illustrations are not included in the reading sample

Table D4 Global Model Goodness-of-Fit Statistics

Illustrations are not included in the reading sample

Table D5 Omnibus Likelihood Ratio Test of Overall Model Effect

Illustrations are not included in the reading sample

Null Hypothesis: Intercept-only model provides sufficient fit to the data

Table D6 Type III Wald Tests of Individual Predictor Effects

Illustrations are not included in the reading sample

Table D7 Parameter Estimates & Exponentiated Odds Ratios (Exp(B))

Illustrations are not included in the reading sample

Note: Threshold terms ([Item8=1], [Item8=2], [Item8=3]) have no odds ratio interpretation; only predictor rows report Exp(B).

Table D8 Covariance Matrix of Parameter Estimates

Rows & Column Order: [Item8=1], [Item8=2], [Item8=3], Item1, Item2, Item3, Item4, Item5, Item6, Item7, Item9, Item10, Item11, Item12

Illustrations are not included in the reading sample

Appendix D Closing Audit Statement

All tables above represent unmodified direct output exported from IBM SPSS 28 after executing the corrected annotated GENLIN syntax provided in Appendix C. No manual rounding, recalculation or value editing was applied to any statistical figures. All coefficient values, test statistics, p-values and model fit metrics match the quantitative interpretation presented in Chapter 4 of the main manuscript. This full raw output archive is retained for peer review audit trail transparency as required by the mixed-methods supplementary material standards.

End of Appendix D

Appendix E:

Three-Stage Manual Thematic Coding Full Mapping Matrix

Coding Framework: Miles & Huberman (1994) three-stage manual qualitative analysis (Open Coding → Axial Coding → Core Thematic Synthesis)

Source Data: 15 anonymised semi-structured interview transcripts (Group A: n=7 High Attitude-Low Competence; Group B: n=5 Balanced EP-CCC; Group C: n=3 BRI Exchange Experience)

Audit Information: Total unique open inductive codes = 67; Inter-coder reliability = 87.2%; All mapping traceable to verbatim interview extracts; Full line-by-line open coding raw record attached as sub-section E1

E0 Preliminary Coding Definition Notes

1. Open Codes: Raw descriptive inductive labels extracted line-by-line from participant speech (no theoretical interpretation at this stage; full 67-code list in E1)

2. Axial Categories: Aggregated higher-order conceptual clusters grouping semantically related open codes, organised around the EP–CCC–BRIAD Diffusion Model’s structural layers

3. Core Overarching Themes: Final synthesised interpretive themes triangulated with N=183 ordinal GLM quantitative results (used for main manuscript Chapter 4.4 analysis)

4. Mapping Matrix Logic: Every open code links to one axial category; every axial category feeds into one unified core theme

E1 Full List of All 67 Unique Raw Open Inductive Codes (Deduplicated)

1. General English class prioritises reading exercises over speaking tasks

2. Exam scores depend heavily on written reading and translation sections

3. I read many English articles about BRI foreign customs

4. I hold positive views of English for cross-border exchange

5. Fear of grammatical mistakes stops me volunteering class discussion

6. No BRI culture role-play activities in compulsory English

7. Teachers rarely organise unscripted group oral talk

8. I know lots of foreign cultural knowledge but cannot speak it out

9. English documentaries increase my interest in BRI partners

10. Nervousness appears when I need to discuss foreign cultures orally

11. All BRI exchange events rely on English communication

12. I want to communicate with overseas youth from Belt countries

13. Oral presentations are voluntary and rarely held

14. Written homework occupies most after-class English learning time

15. Worry about accidentally offending foreigners with wrong words

16. Textbook content only introduces general foreign culture

17. No simulated international negotiation practice in lessons

18. I avoid speaking in mixed-class groups with foreign students

19. Oral assessment weight is very low in semester grades

20. Teachers focus grammar correction rather than dialogue training

21. I absorb Central Asian cultural information via English podcasts

22. Strong willingness to join BRI volunteer programmes

23. Oral anxiety reduces my willingness to sign up exchange activities

24. No scaffolded low-pressure speaking drills in regular class

25. Compulsory English ignores BRI industry English content

26. I rarely get chances to practise spontaneous dialogue

27. Reading materials expand my respect for cultural diversity

28. No teacher guidance on cross-cultural conflict resolution talk

29. Midterm/final tests have almost no oral components

30. Only short prepared speeches are allowed for oral tasks

31. 80% class time used for grammar and reading analysis

32. No interactive multicultural sharing sessions at university

33. I am interested in cross-border youth exchange projects

34. Lack of practice leads to low oral self-confidence

35. General curriculum does not teach negotiation English

36. I dare not apply for international volunteer work

37. Exam pressure pushes teachers to skip oral activities

38. No small-group low-stakes oral workshops on campus

39. Worry about misinterpreting foreign cultural norms when speaking

40. Elective intercultural English has regular role-play sessions

41. Scaffolded speaking tasks reduce my communication anxiety

42. I can freely discuss foreign cultural differences in English

43. Mock BRI negotiation tasks boost oral confidence

44. Balanced reading and speaking allocation in elective modules

45. I can resolve small cultural misunderstandings via English talk

46. Low-pressure oral salon activities available at college

47. Elective ESP course covers BRI trade English scenarios

48. Weekly cross-cultural dialogue group activities held

49. I have no obvious fear of talking about foreign customs

50. Cross-cultural misunderstanding simulation exercises

51. Oral assessment counts for 40% of elective grades

52. Combined input and output eliminates attitude-competence split

53. I actively join English multicultural salons weekly

54. Teachers arrange mixed-nationality online talk activities

55. I apply ELF skills to campus foreign student exchanges

56. University holds monthly BRI English exchange meetings

57. I participate in student cross-cultural dialogue teams

58. Pre-exchange oral training significantly reduces nervousness

59. Many Chinese peers stayed silent during BRI camp group talk

60. I witnessed real cultural misunderstanding events at exchange camps

61. Lack of oral practice restricts youth exchange participation

62. Receptive knowledge cannot solve live communication friction

63. Universal positive views on BRI English among all students

64. Curriculum imbalance creates on-site communication barriers

65. Exchange experience strengthened my ELF learning motivation

66. Exam English fails to match real cross-border dialogue demands

67. Interactive oral training is critical for BRI volunteer performance

E2 Full Three-Stage Coding Mapping Matrix

Illustrations are not included in the reading sample

Core Theme 1: Universal Positive Attitudes Toward Receptive BRI ELF Learning

Axial Category 1: Receptive ELF builds abstract multicultural awareness & positive exchange intent

3,4,9,11,12,21,22,27,33,56,57,63,65

Captures all positive learner perceptions of ELF’s value for BRI exchange, knowledge acquired via silent reading/listening receptive input; maps to Inner Circle Receptive EP + Middle Circle Cognitive-Attitudinal CCC layers.

This theme explains high univariate means for attitudinal survey items in N=183 GLM descriptive statistics.

Core Theme 2: Exam-Centred Curriculum Imbalance Creates the Attitude-Competence Gap

Axial Category 2: Institutional structural neglect of productive oral ELF training

1,2,6,7,13,14,16,17,19,20,25,29,31,32,35,37,38,64,66

Documents systemic teaching and assessment bias toward receptive written skills, absence of BRI-themed oral role-play/negotiation practice; represents the structural root of interrupted diffusion between attitudinal and behavioural CCC in the model’s Middle Circle.

Consistent with non-significant attitudinal covariates after controlling productive EP in Wald test outputs.

Core Theme 3: Cross-Cultural Oral Anxiety Blocks Functional BRI Ambassadorial Participation

Axial Category 3: Affective speaking barriers limiting behavioural CCC & BRIAD engagement

5,10,15,18,23,24,34,36,39,59,60,61,62

Records learner nervousness, fear of cultural/grammatical error, avoidance of spontaneous ELF dialogue; directly inhibits progression from Middle Circle CCC to Outer Circle BRIAD outcomes.

Explains dominant positive Exp(B) value for Item4 (oral dialogue) in full-sample ordinal regression.

Supplementary Balanced Learner Protective Axial Sub-Category (Nested under Core Theme 2 Counterpoint)

Axial Category 4: Elective balanced ELF oral instruction mitigates attitude-competence split

40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,58,67

Captures contrasting positive learning experiences from students with access to low-stakes oral electives; acts as comparative counter-evidence proving oral practice eliminates the attitude-competence gap mechanism in the diffusion model.

E3 Audit Trail & Reliability Statement

1. Open coding: Full line-by-line extraction completed separately for each of the 15 anonymised transcripts (raw extraction trail archived as the preceding E1 section).

2. Axial coding: All 67 open codes manually sorted into the four axial categories above; duplicate semantic labels merged to remove redundancy.

3. Core thematic refinement: Four axial categories condensed into three unified interpretive core themes for mixed-methods triangulation with quantitative GLM results.

4. Inter-coder reliability: A second independent applied linguistics researcher cross-coded 30% of random transcript segments; pairwise coding agreement = 87.2%, above the 85% threshold standard for qualitative ELF research.

5. Traceability: Every open code number in the matrix can be cross-referenced back to the full E1 open code list and original interview transcript verbatim extracts cited in Chapter 4.4 of the main manuscript.

End of Appendix E

Appendix F:

Full Raw SPSS PLUM POM Output (Pilot Sample N=42)

Software: IBM SPSS 28 | Procedure: Ordinal Regression (PLUM)

Model: Standard Proportional Odds Model (POM)

Dependent Variable: Pilot version Item8 (5-point Likert 1–5)

Predictors: Item1, Item2, Item3, Item4, Item5, Item6, Item7, Item9, Item10, Item11, N=42 foreign language undergraduates

Test requested: Brant Test of Parallel Lines; all fit, parameter, correlation tables exported unedited, no manual rounding

Table F1 Case Processing Summary

Illustrations are not included in the reading sample

Dependent Variable: Item8 (Pilot ordinal outcome)

Table F2 Dependent Variable Frequency Distribution

Illustrations are not included in the reading sample

Table F3 Descriptive Statistics (Pilot All Predictors)

Illustrations are not included in the reading sample

Table F4 Model Fitting Information

Illustrations are not included in the reading sample

Null hypothesis: Intercept-only model is sufficient.

Table F5 Goodness-of-Fit

Illustrations are not included in the reading sample

Table F6 Pseudo R-Square

Illustrations are not included in the reading sample

Table F7 Test of Parallel Lines (Brant Assumption Test)

Illustrations are not included in the reading sample

Interpretation: p > 0.05; proportional odds assumption holds for pilot sample.

Table F8 Parameter Estimates (Pilot POM)

Illustrations are not included in the reading sample

Note: Threshold intercepts have no odds ratio interpretation.

Table F9 Correlation Matrix of Parameter Estimates

Rows/Columns order: Item1, Item2, Item3, Item4, Item5, Item6, Item7, Item9, Item10, Item11, Item12

Illustrations are not included in the reading sample

Appendix F Audit Statement

All tables above are unaltered raw outputs exported directly from SPSS 28 PLUM ordinal regression for the pilot N=42 sample. No rounding, recalculation or manual value adjustment was applied. This pilot dataset satisfies the parallel odds assumption (Brant p=0.972), which differs from the full N=183 GLM sample’s variance pattern; comparative divergence is discussed in Chapter 4 and Chapter 5 of the main manuscript. This archive is retained for full peer review audit trail transparency.

End of Appendix F

Appendix G:

Wenjuanxing Cloud Questionnaire Distribution Documentation

Overview: This appendix archives official Wenjuanxing unique survey access links, full platform backend distribution configuration rules, multi-channel standard invitation texts (WeChat group / campus email), data storage ethics protocols and fieldwork timeline records for both pilot (N=42) and full-sample (N=183) questionnaires. All links follow standard wjx.cn official URL format as generated by the platform’s publish module问卷星.

G1 Basic Platform Information

Survey Platform: Wenjuanxing (wjx.cn, China mainstream academic cloud questionnaire system)

Distribution Channels: University disciplinary WeChat class groups, campus international exchange mailing lists, college student union internal notification groups

Data Collection Window: Spring semester 2026 (4 weeks total for full sample; 1 week for pilot)

Backend Restrictions Enabled: Single IP one submission limit, mandatory all-item response, anonymous mode (no phone number collection), no public data sharing switch activated

G2 Pilot Questionnaire (N=42) Full Distribution Records

G2.1 Official Wenjuanxing Access Link

Full pilot survey unique URL (generated via platform publish function):

https://v.wjx.cn/vm/rNXkq6Y.aspx#

Corresponding mini-program WeChat QR Code (WeChat sharing version):

G2.2 Pilot Survey Platform Backend Settings

1. Instrument Version: Pre-revision draft without demographic section (Appendix A1)

2. Mandatory Settings: All 12 Likert items marked required; open-ended feedback text box enabled

3. Response Control: 1 submission per IP address to avoid repeated filling

4. Privacy Mode: Anonymous collection, mobile phone capture function closed

5. Export Format: SPSS .sav file downloadable directly after data collection closed

6. Distribution Scope: Single university foreign language undergraduate WeChat groups only

G2.3 Pilot Short Invitation Text (WeChat Group Version)

Dear students,

I am conducting academic research on university students’ English learning and cross-cultural communication for international exchange. This anonymous questionnaire takes about 8 minutes to finish, all answers are for thesis research only with no personal information collected.

Survey link: https://v.wjx.cn/vm/rNXkq6Y.aspx#

Thank you sincerely for your support of academic research!

G3 Revised Full-Sample Questionnaire (N=183) Full Distribution Records

G3.1 Official Wenjuanxing Access Link

Full formal revised survey unique URL (deployed for multi-university sampling):

https:// https://v.wjx.cn/vm/m5XgCsJ.aspx#

Corresponding mini-program WeChat QR Code (WeChat sharing version)

G3.2 Full-Sample Platform Backend Logical & Distribution Settings

1. Instrument Version: Revised BRI-specific wording + four demographic screening items + optional interview contact text box (Appendix A2)

2. Skip Logic: Automatic exclusion of non-full-time vocational students / overseas long-term residents

3. Response Restriction: One submission per IP; all Likert and demographic items set mandatory

4. Extra Function: Optional blank text field for participants to leave contact info for follow-up semi-structured interviews

5. Data Export: Raw dataset exported as SPSS .sav format, fully compatible with IBM SPSS 28 cumulative logit GLM analysis

6. Sampling Range: Six comprehensive universities across eastern, central and western China; four disciplinary clusters (humanities, social sciences, engineering, natural sciences)

G3.4 Standard Full-Sample Invitation Texts (Two Versions)

Version 1 – Short WeChat Group Sharing Text (High circulation)

Dear students,

We carry out anonymous mixed-methods academic research on Chinese university students’ ELF learning and cross-cultural communication with BRI partner countries. The survey takes 8–10 minutes, all data confidential and only used for linguistic thesis analysis.

If you are willing to join a 25–35 minute online anonymous follow-up interview, you may leave your contact at the bottom page.

Survey link: https:// https://v.wjx.cn/vm/m5XgCsJ.aspx#

Thanks for your honest answers and support!

Version 2 – Formal Campus Email Invitation Text (for college mailing lists)

Subject: Invitation to participate in anonymous academic questionnaire research on ELF and BRI cross-cultural communication

Dear full-time undergraduate and postgraduate students,

This email invites you to take part in a confidential online questionnaire for a mixed-methods ELF intercultural communication study submitted to international applied linguistics journals. The research focuses on the connection between university English learning and cross-cultural dialogue competence within Belt and Road people-to-people exchange contexts.

Key information about this survey:

1. Estimated completion time: 8–10 minutes; all questions are five-point Likert attitudinal items with no right or wrong answers.

2. Full anonymity guaranteed: The Wenjuanxing platform will not collect your name, student ID or phone number automatically. Any contact information you write down at the end is voluntary and separated from survey responses for interview recruitment only.

3. Data usage rule: All collected data will be encrypted and stored offline after download; the cloud platform’s original response records will be permanently deleted to comply with institutional research ethics standards.

4. Eligibility: Full-time mainland Chinese undergraduate / postgraduate students (vocational college students are excluded via screening logic).

Survey access link: https:// https://v.wjx.cn/vm/m5XgCsJ.aspx#

The questionnaire collection window will close on [specific spring semester deadline].

Your genuine feedback provides critical empirical data for this ELF intercultural research. We appreciate your valuable time and participation.

Best regards,

Research Team of EP–CCC–BRIAD Mixed-Methods Study

G4 Data Storage & Ethical Compliance Rules for Wenjuanxing Data

1. Post-collection operation: After survey collection closed, complete raw datasets of pilot and full samples were downloaded as SPSS .sav files, then all original response records stored on Wenjuanxing cloud backend were permanently deleted to eliminate online data leakage risks.

2. Contact data separation: All voluntary interview contact information filled in the optional text box was extracted and saved into a separate encrypted document, completely decoupled from the main quantitative dataset to protect participant anonymity during regression analysis.

3. Storage security: All offline survey datasets, downloaded CSV exports and interview contact files are stored on password-locked local laptop storage with double encryption; no cloud disk uploads of complete participant data.

4. Elimination of identifiable metadata: Platform automatic metadata (IP rough geographic tags, submission timestamps) were removed before importing data into SPSS for statistical modelling to avoid indirect participant identification.

G5 Supplementary Distribution Audit Note

All URLs and invitation texts reproduced above are identical to the actual materials used during 2026 spring semester fieldwork. Screenshot archives of the Wenjuanxing backend publish page, response volume statistics and export operation logs are retained as offline research audit materials for peer review verification upon request.

End of Appendix G

Appendix H

Full Labelled Schematic Structure

Three concentric nested circles arranged centrally: Inner Circle → Middle Circle → Outer Circle

Directional flow: outward sequential diffusion from inner to outer ring.

Illustrations are not included in the reading sample

Figure H1 Diffusion Model

Operationalise Sequential Diffusion Model

1. Inner Circle (Foundational Layer: English Proficiency, EP)

Title: Inner Circle – Multi-Dimensional English Proficiency (EP)

Two internal sub-partitions:

✅ Receptive EP (REP) Sub-dimension
· Reading (Item2)
· Listening comprehension (Item10)

Function: Generates cultural knowledge and cognitive-attitudinal awareness; necessary but insufficient for behavioral cross-cultural competence.

✅ Productive EP (PEP) Sub-dimension
· Oral cross-cultural dialogue (Item4)
· Written cultural expression (Item5)
· BRI Professional ESP (Item6)
· Intercultural negotiation English (Item7)

Function: The only skill set capable of enabling behavioral cross-cultural performance; required for complete unbroken diffusion.

Outward solid arrow extends from Inner Circle toward the Middle Circle.

2. Middle Circle (Mediating Layer: Cross-Cultural Competence, CCC)

Title: Middle Circle – Dual-Dimension Cross-Cultural Competence (CCC)

Two distinct horizontal sub-partitions separated visually:

✅ Cognitive-Attitudinal (CA) CCC

Survey indicators: Item1, Item3, Item9, Item11, Item12

Origins: Primarily built via receptive EP input

Outcome state: Positive cultural openness, abstract multicultural awareness, theoretical recognition of ELF value for BRI cooperation.

❌ Visual Interruption Marker: DASHED BREAK LINE

Annotation on break line: INTERRUPTED DIFFUSION = ATTITUDE–COMPETENCE GAP

Mechanism: When productive EP training is missing, development stalls at cognitive-attitudinal CCC and cannot advance to behavioral CCC.

✅ Behavioral CCC

Measured as the study’s ordinal dependent variable (Item8)

Origins: Requires balanced productive EP practice

Outcome state: Capacity to negotiate and resolve real-time cross-cultural misunderstandings with BRI stakeholders.

Solid arrow proceeds from behavioral CCC toward the Outer Circle. No continuous arrow connects cognitive-attitudinal CCC directly to the Outer Circle.

3. Outer Circle (Outcome Layer: BRI Ambassadorial Diplomacy, BRIAD)

Title: Outer Circle – BRI Ambassadorial Diplomacy (BRIAD)

Practical behavioral outcomes:

· BRI student exchange participation
· Cross-cultural volunteer work
· Simulated & real BRI trade/cultural negotiation
· Youth people-to-people exchange activities

Condition for entry into Outer Circle: Unbroken diffusion chain:

Receptive EP (REP) + Productive EP (PEP) → Integrated Attitudinal (IA) + Behavioral CCC (B) → BRIAD

Supplementary Feedback Loop Annotation

Thin curved feedback arrow from Outer Circle looping back to the Inner Circle:

Annotation: Positive BRIAD experience strengthens motivation for further English learning (reinforcing iterative developmental cycle).

Illustrations are not included in the reading sample

Figure H2 Positive Feedback Loop

Adaptation note: Original Kachru model = static geographic grouping; Present model = dynamic individual learner developmental causal sequence

[...]

Ende der Leseprobe aus 75 Seiten - nach oben

Jetzt kaufen

Titel: English Proficiency, Cross-Cultural Competence and Lingua Franca Participation

Wissenschaftliche Studie , 2026 , 75 Seiten

Autor:in: Dr. Stephen Bodybobton Antwi (Autor:in), Li Xin (Autor:in), Lu Yanling (Autor:in)

Medien / Kommunikation - Interkulturelle Kommunikation

Blick ins Buch

Details

Titel: English Proficiency, Cross-Cultural Competence and Lingua Franca Participation
Untertitel: A Mixed-Methods Analysis of the Attitude-Competence Gap Among Chinese University Learners Within the Belt and Road Initiative Intercultural Context
Veranstaltung: Public Speaking
Autoren: Dr. Stephen Bodybobton Antwi (Autor:in), Li Xin (Autor:in), Lu Yanling (Autor:in)
Erscheinungsjahr: 2026
Seiten: 75
Katalognummer: V1761295
ISBN (PDF): 9783389204672
Sprache: Englisch
Schlagworte: English as a lingua franca; cross-cultural competence; mixed-methods research; ordinal generalised linear model; attitude-competence gap; Expanding Circle intercultural learning
Produktsicherheit: GRIN Publishing GmbH

Arbeit zitieren: Dr. Stephen Bodybobton Antwi (Autor:in), Li Xin (Autor:in), Lu Yanling (Autor:in), 2026, English Proficiency, Cross-Cultural Competence and Lingua Franca Participation, München, GRIN Verlag, https://www.grin.com/document/1761295