Developmental trajectories of English speaking proficiency and perceived intercultural communicative competence among Japanese university students: a latent growth curve analysis

1 Introduction

In today's globalized world, English functions as a lingua franca, connecting speakers from diverse cultural backgrounds and playing a vital role in intercultural communication. The Common European Framework of Reference for Languages (CEFR) highlights the cultural dimensions of language education, emphasizing that the aim of teaching modern languages, including English, is to develop learners’ intercultural competence and related competences (Council of Europe, 2001). Such multicultural language instruction offers learners opportunities to deepen their understanding of both their own culture and others, thereby fostering positive attitudes toward people from different cultural backgrounds (Köşker and Erdoğan, 2020).

Intercultural communicative competence (ICC) generally refers to an individual's ability to interact effectively with people from different cultural backgrounds and to respond appropriately to their perspectives and expectations (Byram, 2021). While learners and teachers of English often prioritize the development of linguistic proficiency, ICC plays an equally important role. English as a Foreign Language (EFL) teachers with strong ICC are better equipped to address cultural differences in the classroom and to manage potential conflicts more effectively (Byram, 2000). Consequently, ICC has attracted increasing attention among scholars and educators in English language teaching over the past decade (Byram, 2021).

ICC is defined as the ability “to interact with people from another social group in another language” (Byram, 2021). The concept encompasses two primary components: communicative competence and intercultural competence. Communicative competence refers to the appropriate use of the target language according to context and includes linguistic, sociolinguistic, discourse, strategic, sociocultural, and social competences (Byram, 2021). Intercultural competence, in contrast, involves the ability to engage with individuals from different cultural backgrounds, even when interacting in one's own language (Byram, 2021).

Byram (2021) conceptualizes ICC as a model aimed at developing the “intercultural speaker,” who can use a foreign language appropriately in intercultural communication. Learners with well-developed ICC are able to communicate more effectively with culturally diverse interlocutors because they possess a deeper understanding of both their own cultural perspectives and those of others

Chen and Starosta (1996). conceptualized intercultural communication competence as the ability to negotiate cultural meanings and communicate both effectively and appropriately while recognizing the multiple identities of interlocutors. Their model comprises three interrelated dimensions: an affective dimension involving intercultural sensitivity, a cognitive dimension involving self-awareness and cultural awareness, and a behavioral dimension involving intercultural communication skills. Although linguistic and message skills constitute an important part of the behavioral dimension, intercultural competence also requires open-mindedness, nonjudgmental attitudes, cultural knowledge, behavioral flexibility, and interaction management.

2 Theoretical framework for the co-development of speaking proficiency and PICC

The present study draws on multidimensional and process-oriented models of intercultural communicative competence. Byram's model conceptualizes ICC as comprising not only communicative competence but also intercultural competence. Similarly, Chen and Starosta (1996) distinguish affective, cognitive, and behavioral dimensions of intercultural communication competence. From these perspectives, linguistic proficiency is an important resource for intercultural communication because it enables learners to express their ideas, negotiate meaning, and respond to culturally diverse interlocutors. However, linguistic proficiency represents only one component of ICC and is not, by itself, sufficient to ensure interculturally appropriate and effective communication.

A process-oriented perspective further suggests that ICC develops through an ongoing and dynamic process involving attitudes, knowledge, skills, internal changes, and external communicative behavior. In Deardorff’s (2006) process model, respect, openness, and curiosity provide a foundation for acquiring cultural knowledge and interactional skills. These knowledge and skills contribute to internal outcomes, including adaptability, flexibility, an ethnorelative perspective, and empathy, which may ultimately lead to effective and appropriate communication and behavior in intercultural situations. Accordingly, the development of ICC is not assumed to follow automatically from improvements in linguistic proficiency. Learners may improve their speaking skills through regular language instruction without experiencing corresponding changes in cultural awareness, empathy, behavioral flexibility, or perceived competence in intercultural situations.

Nevertheless, the two developmental processes may be related. Improved speaking proficiency may allow learners to participate more actively in intercultural interactions, express culturally sensitive responses more precisely, and acquire richer feedback from their interlocutors. These experiences may, in turn, enhance learners’ perceptions of their intercultural communicative competence. Conversely, greater perceived intercultural competence may encourage learners to engage more actively in English-mediated interactions, thereby creating additional opportunities to develop speaking proficiency. Thus, language proficiency and PICC may facilitate one another, but their development may also remain partially independent because each construct is supported by different learning experiences and psychological processes.

This theoretical distinction is particularly important when examining PICC rather than objectively demonstrated ICC. PICC reflects learners’ generalized perceptions of their ability to communicate effectively and appropriately in intercultural situations. Such perceptions need not correspond directly to objectively assessed speaking proficiency and may therefore show a different pattern of change over time. The present study consequently examines whether the two constructs show parallel developmental trajectories without assuming that improvement in one construct necessarily produces improvement in the other.

3 Previous empirical studies

The relationship between English proficiency and intercultural communicative competence (ICC) remains a subject of debate. Hismanoglu (2011) reported no significant differences in the overall acquisition of ICC between ELT students with lower and higher levels of language proficiency. However, qualitative findings suggested a more nuanced picture: students with higher proficiency tended to provide more appropriate and contextually suitable responses in communicative situations compared to their lower-proficiency peers. Overall, the results indicated that while general ICC levels may not differ significantly, higher language proficiency appears to be associated with more acceptable and effective performance in intercultural communicative contexts.

A study by Iino (2020) examined the relationship between perceived intercultural communicative competence and English speaking ability. In this study, Japanese university students who learned English were assigned weekly tasks in which they conducted interviews via online conversation platforms with English speakers living overseas. As a result, the learners showed improvement in their speaking proficiency as well as their tolerance toward other cultures. The findings suggest that activities in which learners interact with English speakers as conversation partners and exchange opinions on current issues can contribute to the development of intercultural communicative competence, indicating that speaking ability and intercultural communicative competence may develop together.

Furthermore, Iino and Tanaka (2025) showed a positive correlation between English speaking proficiency and perceived intercultural communicative competence among Japanese university students. They developed a questionnaire consisting of 30 items targeting speech and attitudes in intercultural communication context. As a result of factor analysis, they identified four factors: adjusting mutual understanding while being mindful of the human relationship, speaking with consideration for the interlocutor's cultural perspective, showing interest in and empathy toward the interlocutor's opinions, and expressing one's opinions logically while being considerate of the interlocutor.

More recently, Wang et al. (2025) examined 223 Chinese pre-service teachers and found that self-reported English proficiency was positively associated with self-reported ICC. English proficiency was also identified as a significant predictor of ICC in their cross-sectional analysis. Although these findings support a between-person association between language proficiency and ICC, the cross-sectional design does not establish whether within-person changes in the two constructs are developmentally linked.

Taken together, previous studies suggest that language proficiency and ICC or PICC are positively associated at the between-person level and that both may improve in educational programs involving language instruction and intercultural interaction. However, these findings do not establish whether within-person improvements in speaking proficiency are accompanied by corresponding changes in PICC. Cross-sectional correlations cannot distinguish developmental coupling from stable individual differences, while intervention studies often incorporate multiple educational components, making it difficult to determine whether changes in language proficiency and PICC are developmentally linked. Consequently, it remains unclear whether speaking proficiency and PICC follow parallel trajectories within the same learners over time. The present study addresses this gap by using a bivariate latent growth curve model to examine their mean trajectories, individual differences in change, and covariation between their developmental parameters.

4 Objectives and research questions

The present study exploratorily examined the developmental trajectories of English speaking proficiency and PICC over one academic year and whether changes in the two constructs were associated. To track change over time, we assessed Japanese university students at three measurement occasions. First, we examined whether English speaking proficiency and PICC show average change over one academic year. Second, we examined whether individuals differed significantly from one another in their rates of change over time for English speaking proficiency and PICC. Third, we examined whether the initial level of each construct was associated with its subsequent rate of change. Fourth, we examined whether changes over time in English speaking proficiency were related to changes over time in PICC.

• RQ1: Did English speaking proficiency and PICC show average change over one academic year?
• RQ2: Were there individual differences in the rates of change in English speaking proficiency and PICC?
• RQ3: Were the initial levels of the two constructs associated with their subsequent rates of change?
• RQ4: Was change in English speaking proficiency associated with change in PICC?

5 Latent growth curve modeling (LGCM)

Researchers have employed Latent Growth Curve Modeling (LGCM) within a structural equation modeling (SEM) framework to examine changes in individuals’ attributes over time. Compared with traditional regression techniques and mean comparisons, LGCM addresses key methodological limitations in the analysis of panel data. Rather than treating repeated measurements as static observations at separate time points, LGCM conceptualizes them as part of a developmental process unfolding over time (Coyne and Downey, 1991). Moreover, when individual changes follow a nonlinear trajectory, LGCM provides an appropriate approach for capturing the complexity of such patterns (Willett and Sayer, 1994).

One important model specification is the latent basis model, which treats the pattern of change itself as a latent construct (Meredith and Tisak, 1990). Unlike models that impose a prespecified growth pattern, the latent basis model estimates the trajectory directly from the data. This flexibility allows researchers to detect differences in the growth processes across observed groups. Notably, at least three time points are required to test a linear trend (Felt et al., 2017).

As Elahi Shirvan et al. (2021) explain, LGCM estimates a growth curve that represents change over time across participants. It yields an average intercept (initial level) and an average slope (rate of change) for each latent variable, while also estimating individual variability around these parameters. Furthermore, LGCM enables the examination of covariances between latent growth factors, thereby allowing researchers to investigate the co-development of constructs over time. In the present study, LGCM was used to examine whether changes in English speaking proficiency were associated with changes in PICC across three measurement occasions.

Specifically, we modeled the trajectories of two time-varying constructs—English speaking proficiency and PICC—across three time points. We estimated the mean intercepts and slopes for each construct, along with their variances, to capture individual differences in initial levels and rates of change. Finally, we analyzed the covariance between the growth factors of English speaking proficiency and PICC to explore their longitudinal co-development.

According to Preacher (2018), a primary advantage of LGCM is its ability to model change parameters (intercepts and slopes) as random effects. This approach allows researchers to estimate the means, variances, and covariances of individual differences in growth trajectories. Because LGCM is a special case of SEM, it retains the broader advantages of SEM, including overall model fit assessment, the modeling of latent variables, and the examination of predictors and outcomes of change. In addition, LGCM handles missing data effectively under the missing-at-random assumption, and it does not require measurements to occur at identical or equally spaced time intervals.

6 Methods

6.1 Participants

The participants were 64 Japanese undergraduate students from a university in Tokyo (31 men, 33 women; mean age = 19.94 years, SD = 0.91; average TOEIC score = 558.61, April 2024). They attended regular English classes. They were from the academic seminars of three professors. The professors’ areas of specialization were critical thinking, intercultural communication, and psychology. In each seminar, they studied English to develop effective and practical English communication competence. In addition, they studied discussions of current issues, speaking strategies, and even statistical techniques to conduct a survey. Participants from three academic seminars were recruited by their professors and received course credit for their participation.

6.2 Procedures

The participants answered questionnaires assessing perceived intercultural communicative competence and took the PROGOS speaking test three times: at the beginning (April 2024), middle (July 2024), and end (January 2025) of the academic year. Therefore, we could focus on the changes of participants throughout the year, and this longitudinal design allowed us to examine whether within-person changes in the two constructs occurred together. The instructors distributed links to web-based questionnaires to the students, who completed the surveys using their own computers or smartphones. On the first page of the questionnaire, written in Japanese, participants were informed that their participation was voluntary and anonymous. They provided informed consent to participate in the study and to allow the use of their data for research purposes. In addition, the instructors distributed links to the PROGOS speaking test, which participants completed using their own computers or smartphones.

6.3 PROGOS speaking test

The participants took the PROGOS English Speaking Test, an online test developed by the PROGOS Corporation that uses AI-based scoring. The test takes approximately 20 min and consists of five parts: an interview (responding to short questions), reading aloud (reading displayed sentences), a presentation (giving a speech based on a specified topic and three key points), a presentation using graphs and charts (describing visual information), and a role play (engaging in several turns of interaction according to assigned roles). After completion, test-takers receive score reports aligned with CEFR-J (an adapted version of the international Common European Framework of Reference for Languages, specifically redesigned for Japanese learners of English), and administrators can access learners’ test histories and score lists online. The overall proficiency level is reported on nine levels, ranging from Pre-A1 to B2 High and Above, which were converted into a nine-point scale to examine the relationship with perceived intercultural ability (from Pre-A1 = 1 to B2 High and above = 9). Its reliability is supported by data from approximately 420,000 test-takers, which provide a large-scale picture of English-speaking proficiency among Japanese learners. Its validity is supported by its assessment of six CEFR-based dimensions essential for communication in business contexts: range, accuracy, fluency, interaction, coherence, and phonology.

6.4 Perceived intercultural communicative competence (PICC)

The Japanese version was developed by Iino and Ito (2023), and was based on Miyamoto and Matsuoka (2000). Thirty items were used to prompt reflection on speech behavior in intercultural communication contexts (α = .91 at Time 1, α = .92 at Time 2; α = .91 at Time 3). The instrument was administered to the participants with the following instructions: “Please imagine a situation in which you are speaking English with someone who lives in a foreign country. Choose the option that best applies to you for each question.” Responses were recorded on a five-point Likert scale: 1 = almost never, 2 = seldom, 3 = sometimes, 4 = often, 5 = almost always. This scale is assumed to consist of four factors: adjusting mutual understanding while being mindful of the human relationship (e.g., making an effort to speak using words, phrases, and expressions that are easy for the interlocutor to understand), speaking with consideration for the interlocutor's cultural perspective (e.g., trying to speak while taking into account the other person's cultural background, lifestyle, and way of life), showing interest in and empathy toward the interlocutor's opinions (e.g., adopting an attitude of showing as much interest as possible in topics related to the country or region where the interlocutor lives or comes from), expressing one's opinions logically while being considerate of the interlocutor (e.g., talking about one's own views while using expressions that respect the interlocutor's ideas and position). The questionnaire assessed generalized self-perceptions of communicative behaviors and intercultural orientations in imagined English-mediated interactions. The communicative component included behaviors such as using comprehensible expressions, adjusting communication to facilitate mutual understanding, and expressing opinions logically. The intercultural component included considering the interlocutor's cultural background and showing interest in and empathy toward the interlocutor's views.

7 Results

The average scores of the PROGOS speaking test were Time 1: M = 5.45, SD = 1.64; Time 2: M = 5.90, SD = 1.34; Time 3: M = 5.75, SD = 1.40. The average scores of perceived intercultural communicative competence (PICC) were Time 1: M = 3.67, SD = 0.48; Time 2: M = 3.63, SD = 0.53; Time 3: M = 3.75, SD = 0.48.

Table 1 shows the correlations between PROGOS and PICC. At Time 2, PROGOS and PICC were positively correlated (r = .31, p < .05). However, the other relationships were not statistically significant.

Table 1. Correlation matrix of PICC and PROGOS.

p < .05.

p < .01.

The bivariate latent growth curve model showed acceptable model fit to the data, χ²(7) = 7.58, p = .37, CFI = .995, TLI = .989, RMSEA = .035 [90% CI = .00,.16], SRMR = .054. On average, participants’ PROGOS scores increased slightly over time, b = 0.17, p = .037, whereas their PICC did not show a significant change, b = 0.03, p = .39. Figure 1 shows the individual-level and mean growth trajectories.

Figure 2 shows the variance and covariance estimates of the latent intercepts and slopes. The variances of the latent intercept factors were statistically significant for both PROGOS (Var = 1.20, p = .004) and PICC (Var = 0.13, p = .04), indicating substantial individual differences in initial status for both constructs. In contrast, the variances of the latent slope factors were not statistically significant for either PROGOS (Var = −0.17, p = .43) or PICC (Var = 0.003, p = .94). These results indicate that, although the average rate of change differed from zero for PROGOS, individuals did not differ significantly from one another in their rates of change over the three time points.

The covariance between the initial level (i1) and the growth rate (s1) of PROGOS was not significant (Estimate = 0.19, p = .39), indicating that there was no relationship between initial English speaking proficiency and the rate of growth. The results showed that participants who started with higher English speaking proficiency did not tend to grow faster or slower afterward. In addition, the covariance between the initial level (i2) and the growth rate (s2) of PICC was not significant (Estimate = 0.003, p = .94), indicating that there was no relationship between the initial level and growth rate. The results showed that participants who started with higher perceived intercultural competence did not tend to grow faster or slower over time.

Regarding covariances among growth factors, the covariance between the PROGOS slope and the PICC slope was not statistically significant (Estimate = 0.014, p = .43), indicating no evidence that changes in language proficiency and perceived intercultural competence were systematically related. The covariance between the intercept factors showed no statistical significance (Estimate = 0.16, p = .07). Therefore, the results do not provide evidence that participants with higher initial English speaking proficiency tended to report higher initial levels of perceived intercultural competence.

8 Discussion

The bivariate latent growth curve model showed an acceptable fit to the data. The parallel process latent growth curve model revealed that although English speaking proficiency demonstrated a modest overall upward trend across the three occasions, perceived intercultural communicative competence did not show a significant mean growth trajectory (RQ1).

The variance of the latent intercepts and slopes showed that even though substantial individual differences in initial status for English speaking proficiency and perceived intercultural competence were significant, individuals did not differ significantly from one another in their rates of change over time (RQ2). While substantial individual differences in initial levels of both constructs were shown, individuals did not differ from one another in their rates of change over time. The non-significant slope variances may reflect the limited number of measurement occasions and the relatively homogeneous developmental trajectories across participants.

The covariance between the initial level and the growth rate of English speaking proficiency or perceived intercultural competence was not significant, indicating that there was no relationship between initial status and the rate of growth for both constructs (RQ3). The results showed that participants who started with higher proficiency or perceived competence did not tend to grow faster or slower afterward.

Importantly, in addition that the covariance between the intercept factors of English speaking proficiency and perceived intercultural competence showed no significance, no significant association was found between the growth rates of the two constructs, indicating that improvements in English speaking proficiency were not accompanied by corresponding changes in perceived intercultural competence within the observed period (RQ4). These findings suggest that the two constructs did not show evidence of coupled change over the observed period, although the nonsignificant association should not be interpreted as evidence of complete developmental independence.

These findings can be interpreted in light of the multidimensional and process-oriented models of ICC described in the theoretical framework. Although linguistic proficiency provides an important resource for intercultural communication, Byram’s (2021) and Chen and Starosta’s (1996) models suggest that ICC also involves cultural knowledge, attitudes, intercultural sensitivity, behavioral flexibility, and interactional skills. Similarly, Deardorff’s (2006) process model proposes that effective and appropriate intercultural communication develops through the interplay of attitudes, knowledge, skills, and internal changes, including adaptability and empathy. Accordingly, improvement in speaking proficiency may not be accompanied by an immediate change in PICC if these additional components develop through different experiences or at different rates. The increase in PROGOS scores without a corresponding significant increase in PICC is therefore consistent with the theoretical proposition that linguistic and intercultural development are related but partially distinct processes.

The present findings do not necessarily contradict previous cross-sectional evidence showing a positive between-person association between language proficiency and ICC (Wang et al., 2025). Learners with higher language proficiency may generally report higher ICC, while within-person changes in speaking proficiency and PICC may not occur simultaneously over a single academic year.

The results of the present study imply that gains in linguistic proficiency alone may not be sufficient to foster growth in perceived intercultural communicative competence. Consistent with multidimensional and process-oriented models of ICC, language instruction may need to provide explicit opportunities to develop cultural awareness, empathy, behavioral flexibility, and reflection on intercultural experiences. Regular English classes and seminar activities may improve students’ language skills; however, PICC may not develop automatically unless intercultural experiences and structured opportunities for reflection are intentionally incorporated into instruction.

8.1 Limitations

We explicitly acknowledge that the nonsignificant association may reflect limited statistical power and the small estimated variability in the slopes. Therefore, future study needs replication with a larger, more diverse sample, and additional measurement occasions.

Although the questionnaire demonstrated high internal consistency at each measurement occasion, longitudinal measurement invariance was not formally examined. Therefore, it remains uncertain whether the scale measured PICC in an equivalent manner across the three time points. Future research with a larger sample should examine whether the measurement structure remains invariant over time and assess possible repeated-testing effects.

The present study assessed self-perceived rather than objectively demonstrated intercultural communicative competence. Participants may have overestimated or underestimated their competence, and their questionnaire responses may not correspond directly to their actual performance in intercultural interactions. Future studies should combine self-report measures with behavioral assessments, standardized interaction tasks, and evaluations by trained observers.

Several limitations concerning the measurement of PICC should be acknowledged. We did not systematically assess participants’ prior intercultural communication experiences, and participants may have differed in the experiential basis and accuracy of their self-assessments. In addition, the broadly worded instruction may have led participants to imagine different interlocutors and situations (the interlocutor (e.g., professor, close friend, etc.), the interlocutor's English proficiency (e.g., L1 English speaker, ESL learners, or EFL learners), nationality, gender, age, or cultural backgrounds (e.g., European speaker of English versus Asian English speaker). The scale should therefore be interpreted as measuring generalized self-perceptions rather than actual competence in a standardized intercultural interaction. It is also possible that linguistic proficiency must exceed a certain threshold before learners can fully demonstrate interculturally appropriate communication behaviors; however, the present study was not designed to test this possibility. Future studies should use standardized scenarios and combine self-report measures with behavioral assessments and observer ratings.

9 Conclusion

The latent growth curve modeling results showed a modest average increase in English speaking proficiency but no significant average change in PICC over the academic year. Furthermore, we found no evidence that changes in English speaking proficiency were systematically associated with changes in PICC during the observed period. These findings are consistent with multidimensional and process-oriented models of ICC, which suggest that linguistic proficiency is an important resource for intercultural communication but may not be sufficient, by itself, to promote growth in perceived intercultural communicative competence. Language education may therefore benefit from incorporating targeted intercultural experiences and structured opportunities for reflection alongside speaking instruction.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Jiayi Wang, De Montfort University, United Kingdom

Reviewed by: J. Mary Jennifer, Vellore Institute of Technology, India

Fatmawati Djafri, Gadjah Mada University, Indonesia