The influence of networks of general trust on willingness to communicate in English for Japanese people

Research on second language (L2) communication has sought to identify factors shaping attitudes toward L2 communication, with willingness to communicate (WTC) conceptualized as a readiness to enter into discourse in the L2. Prior studies across Canadian, Chinese, Iranian, Japanese, and Korean learners link WTC to frequency of L2 communication and to psychological factors (e.g., confidence, Big Five traits, attitudes, motivation) as well as contextual and situational classroom factors (e.g., teacher support, student cohesiveness, task orientation). An ecological approach highlights the role of interpersonal trust-related factors in fostering WTC. This study focuses on general interpersonal relationships, specifically general trust, which is salient in Japan’s social structure. Prior work (Yamagishi) suggests Japanese interpersonal networks are less flexible than in Western societies, with fewer opportunities to build new relationships and generally lower trust. General trust, an individual disposition toward others’ goodwill, is expected to support WTC in English by facilitating broader social networks. A previous regression study (Ito) found general trust positively influenced WTC in English among Japanese people. However, the network structure among trust components and their distinct effects on WTC remain unexplored. This study aims to reveal the network structure of general trust and its links to WTC using psychological network analysis, which can model multiple interacting variables and identify central (hub) variables potentially useful for interventions. To examine generalization, two groups differing in language experiences (university students and social survey adults) were analyzed.

- WTC in L2 has been linked to actual communication frequency and is influenced by psychological factors (confidence, personality, attitudes, motivation) and situational/classroom factors. - Studies in Korea and Iran show classroom social climate (teacher academic/emotional support, mutual respect), teacher support, student cohesiveness, and task orientation influence English WTC through basic psychological needs and environment. - Situational antecedents of WTC include situation cues (teacher, class, peers, activity, topic), situation characteristics (support, cooperation, objectives), and dimensions (negativity, positivity, duty). - Ecological classroom research emphasizes interpersonal trust (e.g., teacher support involving help, friendship, trust; student cohesiveness involving mutual support) as relevant to WTC. - General trust in Japan: Japanese society characterized by less flexible networks and lower general trust; lower trust can constrain social interaction opportunities, potentially reducing WTC. - Prior regression (Ito) found general trust positively influences English WTC among Japanese learners, but did not reveal the internal network structure of trust variables or their centrality. - Network analysis in psychopathology shows value in modeling symptoms as interacting nodes; compared to latent variable models, network models estimate unique variance and can suggest causal pathways. Compared with SEM/mediation, network approaches flexibly include many variables without strict prior structural assumptions.

Design: Cross-sectional survey with psychological network analysis and regression. Participants: Total N=1,362 Japanese participants comprising 761 university undergraduates from six universities in Tokyo, Kanagawa, and Akita (372 men, 384 women, 5 others; mean age 19.37, SD 1.20) enrolled in mandatory English speaking/listening and reading/writing courses taught by native or Japanese English-speaking instructors; and 601 adults from a social survey panel in Tokyo (284 men, 317 women; mean age 39.65, SD 11.00). Language experience indicators: - Chance of communicating with English speakers (1=Not at all to 5=Every day; dummy-coded 0 for 1, 1 for 2–5) differed by group, with higher chances in the social survey group (χ²(1)=15.68, p<0.001). - Experience staying in foreign countries (1=Never to 5=>6 months; dummy-coded 0 for 1, 1 for 2–5) was higher in the social survey group (χ²(1)=10.79, p<0.01). - Reported past-month communication outside class: 120/761 students; 271/601 social survey participants. - Reported opportunity to stay abroad: 363/731 students; 353/601 social survey participants. Procedure: - University students completed paper or web-based questionnaires during classes after informed consent; participation was voluntary and anonymous. - Social survey participants completed an online questionnaire via Fastask (Just System); participation was voluntary and anonymous; compensation provided. - Ethical approval: Research Committee of the Center for English as a Lingua Franca at Tamagawa University; procedures followed relevant guidelines. Measures: - General Trust Scale (Yamagishi & Yamagishi, Japanese version): 6 items rated 1–5 (strongly disagree to strongly agree). Items: 1. Most people are trustworthy; 2. Most people will respond in kind when they are trusted by others; 3. Most people are trustful of others; 4. Most people are basically honest; 5. I am trustful; 6. Most people are basically good and kind. Reliability: α=0.86 (all), 0.84 (students), 0.89 (social survey). - Willingness to Communicate in English (McCroskey; Japanese version): 12 items across 4 contexts (dyads, small groups, large meetings, public speaking) and 3 receiver types (strangers, acquaintances, friends), rated 1–5 (not at all to very). Reliability: α=0.96 (all), 0.95 (students), 0.97 (social survey). Analytic approach: - Preliminary regression analysis: Linear regression of overall general trust on WTC performed for all participants and separately by group. - Network estimation: Partial correlation network estimated using graphical LASSO regularization with EBIC model selection to identify edges (unique associations controlling for all other nodes). Implemented in R (qgraph, glasso). Separate networks estimated simultaneously for university students and social survey groups (correlation matrices computed per group; dplyr for data prep). Nodes: six trust items (T1–T6) and WTC. - Centrality metrics: node strength (sum of absolute edge weights), closeness (inverse of summed shortest path lengths), betweenness (number of shortest paths passing through node), expected influence (signed strength), and Zhang signed clustering coefficient. - Bootstrap and stability: Nonparametric bootstrapping (bootnet; nBoots=2500, nCores=8) to obtain confidence intervals for edge weights; case-dropping bootstrapping to compute correlation-stability (CS) coefficients for centrality indices. R version referenced 4.0.2 in figure caption. Visualization: Network graphs for all participants and by group; centrality plots; bootstrapped edge-weight CI plots; centrality stability plots.

- Regression results: General trust positively predicted WTC in English. - All participants: β=0.24, p<0.01. - University students: β=0.18, p<0.01. - Social survey: β=0.31, p<0.01. - Network analysis (all participants): Four trust variables had direct positive effects on WTC: T1 (Most people are trustworthy), T3 (Most people are trustful of others), T4 (Most people are basically honest), T5 (I am trustful). T2 (Most people will respond in kind when trusted) and T6 (Most people are basically good and kind) affected WTC indirectly through other variables. - Group-specific networks: - University students: T1, T2, T4, T5 positively associated with WTC; T6 negatively associated with WTC. - Social survey: T1, T3, T4, T5 positively associated with WTC; T2 negatively associated with WTC. - Notably, T2 showed opposite effects across groups (positive for students, negative for social survey participants). - Centrality: Strength, closeness, and expected influence patterns were similar across groups; T1, T3, and T4 showed high centrality values. Variable T1 had a strong positive connection with WTC and high centrality, suggesting a hub role. - Bootstrapping: - Many edge weights’ bootstrapped CIs overlapped, indicating caution in rank-order interpretation. - Highest edge weight among trust variables: T3–T4 (positive); highest WTC-related edge weight: T1–WTC. - Centrality stability (CS, cor=0.7): strength=0.75, closeness=0.75, expected influence=0.75 (reliable); betweenness=0.33 (unreliable). - Group differences in language experiences: Social survey group reported higher chances to communicate with English speakers (χ²(1)=15.68, p<0.001) and more foreign country experience (χ²(1)=10.79, p<0.01) than students.

The study confirmed that general trust is positively associated with WTC in English among Japanese participants, replicating prior regression findings and extending them with network analysis to reveal the internal structure among trust components. Network modeling showed that specific components of general trust (notably T1, T3, T4, T5) directly promote WTC, while others exert indirect or even negative effects depending on subgroup. Centrality analyses identified T1, T3, and T4 as central nodes with strong potential leverage on WTC through the network, providing targets for interventions (e.g., classroom practices that build the belief that most people are trustworthy). Group comparisons indicated that the reciprocal trust belief (T2) operates differently by context: it was positively associated with WTC among university students but negatively among adults in the social survey, potentially reflecting differing real-world communication experiences (more opportunities and possibly negative experiences among adults). Bootstrapping supported the reliability of strength, closeness, and expected influence indices, while betweenness was unstable. Overall, network analysis offered richer insights than regression by identifying central variables and sign patterns of edges, informing practical strategies to enhance WTC through fostering specific facets of general trust.

This study elucidated the network structure of general trust and its effects on willingness to communicate in English among Japanese people. Network analysis identified which trust components directly affect WTC and which function as central hubs (especially “Most people are trustworthy”), offering actionable targets for interventions in language education. By revealing interactions among trust variables, the findings deepen understanding of how general trust shapes L2 communicative attitudes and can guide intervention studies to enhance WTC. Future research should test generalizability across cultural contexts with differing baseline trust and social network flexibility and further examine how experiences influence the operation of specific trust components.

- The sample was limited to Japanese participants; findings may not generalize to Western or other cultural contexts where general trust and social network flexibility differ. - Betweenness centrality showed low stability (CS=0.33), limiting interpretability of that specific centrality index. - Cross-sectional design precludes causal inference; network edges suggest associations rather than confirmed causality. - Self-report measures may be subject to response biases; communication opportunity measures were coarse (dummy-coded).