Content area
In computer-assisted language learning (CALL), a variety of studies have explored the use of Virtual Exchange (VE) for shaping learners' cultural competence from the perspective of sociolinguistics. However, few studies have examined learners' psycholinguistic factors in this context. Since based on the dynamic complex system, students' psycholinguistic factors act independently and in conjunction with each other in any language learning context. Thus, by applying stimulus-organism-response theory (S-O-R), we explored these inter-correlations in this context. In this case, we used Interactive, Constructive, Passive, and Active activities in CALL (ICAP), which have not been integrated into the field as stimuli to shape learners' online engagement as organisms and approaches to VE as a response. Accordingly, we integrated VE tasks into their language and cultural exchanges with Cypriot and Irish students in their respective partner universities based on ICAPCALL. The ICAPCALL and Students' Approach to Virtual Exchange (SAVE) were validated through partial least square modeling (PLS-SEM) in the VE. Moreover, the serial mediation analysis showed that learners could establish knowledge individually or collaboratively, influencing behavior and cognitive engagement to ask questions and exchange information to solve complex learning problems, stimulating a VE deepening approach. Moreover, the Necessary Conditional Analysis (NCA) revealed that passive and active learning activities were among the necessary conditions in shaping learners' deep approach to VE; however, a further level of them could not increase their deep approach to it. Based on these findings, the study introduces a new psycholinguistic framework to the VE, namely as a SAVE, and language learning activities to the CALL literature, namely as an ICAPCALL, and provides a new interconnectedness of learners' dynamic complex systems in the VE and highlights the importance of teachers maintaining a balance between collaborative and individual learning activities in this context to shape students' deep approach.
Introduction
In the past two decades, Virtual Exchange has gained traction, becoming a distinct subfield of CALL as well as a central component of intercultural communication. O’Dowd (2012) describes VE as ‘the application of online communication tools to bring together classes of language learners in geographically distant locations to develop their foreign language skills and intercultural competence through collaborative tasks and project works’ (p. 340).
Many studies have demonstrated that VE can lead to the development of second and foreign languages skills (Luo & Gui, 2019), Intercultural Communicative Competence (ICC, O’Dowd, 2013), and higher-order thinking skills (Ou Yang et al., 2023). In recent studies, various simplified and technologically-driven strategies have been analyzed to uncover the role of language teachers' VE procedures in developing students' ICC and language skills (Chen et al., 2013; Wu, 2021). Despite this, the analyses of these procedures were time-consuming, and instructors were found to be inconsistent when implementing what was specified in their plans (Chi et al., 2018). Furthermore, most of these studies fail to examine teachers' VE procedures from the perspective of language learners.
Other studies have dealt with the frequency of implementing VE to enhance language learners' language skills, subskills, or ICC (e.g. Luo & Gui, 2019). However, no information was provided regarding how language teachers integrated and implemented VE tasks to promote students’ ICC. A third research strand examined was teachers' cultural awareness and proficiency (e.g., Technological Pedagogical Content Knowledge, TPACK, Rienties et al., 2020) while implementing VE. Nevertheless, the surveys which informed such studies only focused on: (a) an overall index of teachers' cultural awareness or technology; (b) pedagogy literacies in developing language learners' ICC; or (c) language use as an outcome variable; all of which are insufficient to provide a qualitative understanding of VE. According to Backfisch et al. (2021), technology implementation quality deepens in terms of the manner in which it is integrated and reshaped by learning activities, as well as with regard to the level to which teaching quality is incorporated in task-specific strategies (e.g., cognitive, emotional and behavioral activation). In order to isolate these gaps, this study takes a step further and shifts the view from the frequency of VE usage (quantity) to how it can be integrated (quality) based on ICAP (Antonietti et al., 2023) taxonomies and taking into account learners' perspectives.
In addition, most studies have found that VE has successfully developed learners' ICCs, language skills, or subskills, indicating its success from both a sociolinguistic and a language teaching perspective. However, unlike face-to-face settings, VE's ecosystem has unique affordances and constraints that may influence the learners’ psychological factors. According to Müller-Hartmann (2005), internet-mediated intercultural exchange is both a highly complex social activity and pedagogically challenging, as the “complex learning environments of telecollaborative exchanges” (Kurek, 2015, p. 27) and “the complexity of the actual telecollaborative classroom” (Grau & Turula, 2019, p. 110) culminate in “instability and unpredictability” due to a host of different elements involved (i.e. tasks, participants, materials; Kurek, 2015). Issues such as failures in communication, technological difficulties and tensions, conflicts, and power issues resulting from differences in expectations, motivations, and cultural background add to the complexity of the situation. Due to the complexity of VE practices, it is clear that psychological traits like motivation, engagement, and self-regulation are key to their success. In this way, students will be able to cope better with the difficulties and complexity involved, which will improve their academic performance. For students to persevere and succeed, they must develop cognitive, affective, and behavioral factors. These factors include motivation, self-regulation, engagement, L2 grits, and emotional which are among the dynamic complex system in second language learning (SLA, Freeborn et al., 2022; Larsen-Freeman & Cameron, 2008) but have rarely been examined in VE. To put it simply, VE has received relatively little attention as a successful tool from the psycholinguistic perspectives. In a recent integrative review, leading scholars recommended focusing on aspects of VE other than language skills and ICC (Colpaert, 2020; Luo & Yang, 2018; O'Dowd, 2021). In order to bridge this gap and implement these recommendations, we not only have shifted the focus from the quantity of VE integration to its quality but also investigated how this quality integration can influence language learners' online engagement and their learning approaches to VE. The reason for this is that Antonietti et al. (2023) found that original ICAP can significantly contribute to learners' cognitive and behavioral skills, which are crucial to understanding in a variety of contexts for digital learning. The role it plays in shaping language learners' cognitive, emotional, and behavioral engagement and their deep and surface approach to VE has been overlooked. To fill this void, we address the following research questions:
RQ1
What are the factorial structures regarding the validity and reliability of constructive, passive, and active activities in CALL (ICAPCALL), as well as students' approaches to virtual exchange (SAVE) in the Spanish EFL context?
RQ2
Does interactive, constructive, passive, and active activity in CALL (ICAPCALL) shape learners’ online engagement and learning approaches to virtual exchange?
RQ3
Does language learners' online engagement predict their deep and surface approaches to virtual exchange?
RQ4
Does language learners’ online engagement mediate the relationship between language learners’ Interactive, Constructive, Passive, and Active activities in CALL (ICAPCALL) and their approaches to virtual exchange?
RQ5
In the presence of Interactive, Constructive, Passive, and Active activities in CALL (ICAPCALL) and learners' online engagement in virtual exchange, what conditions are necessary to shape students' approaches to virtual exchange (SAVE)?
Literature review
ICAP framework
A majority of surveys conducted during the early stages of ICT integration in education examined whether teachers and students used them frequently, felt comfortable implementing them in education, or were satisfied after using them (Rahimi, 2023). Despite their usefulness, these measures cannot be used to demonstrate the effectiveness of ICT in learning and teaching. (Antonietti et al., 2023). Also, neither the frequency of use of target technology nor attitudes towards technology use in the classroom provide helpful information about the pedagogy underpinning technology integration. This in turn prevents the exploration of the implications of target ICT for teaching and learning in the long run (Fütterer et al., 2023). This gap was filled by Antonietti et al. (2023), who developed the ICAP framework to elaborate on learning activities based on Chi et al.’s (2018) and Chi’s (2009) definitions of cognitive engagement in the learning process. Technology implementation in teaching and learning is only beneficial if it is in harmony with the pedagogical purposes of the teaching and learning activity, and this is particularly true in the case of VE. This is due to the fact that its integration into language learning requires a critical examination of the pedagogical principles that underlie its application. Despite leading experts' recommendations to address this gap (Colpaert, 2020; O’Dowd, 2021), previous studies have not addressed this issue in VE. For this reason, the ICAP framework is applied in this study with a view towards determining the level of students' online engagement in activities, supported by how those activities are implemented by their teachers. This provides insights into the quality integration of VE. Taking into account the study context, the ICAP components were operationalized as follows:
Passive (PS): VE activities in which teachers present content with predefined knowledge and learners merely have receptive roles.
Active (AL): Activities that provide learners with a hands-on opportunity to apply their newly acquired knowledge in VE.
Constructive (CL): Individual activities in VE that allow students to develop new knowledge and establish new links between knowledge elements (e.g., creating concept maps, comparing information, solving problems), enabling them to move beyond the institutional materials provided by the instructors.
Interactive (IL): VE activities that involve the interaction and cooperation between learners and their partners in the construction of their knowledge based on their previously acquired one and the information they receive from their partners.
Online engagement
The last two decades have seen a rise in the importance of learning engagement for academic development and achievement (Fredricks et al., 2016; Skinner & Pitzer, 2012). According to Reeve (2012), engagement is the extent to which students actively participate in learning activities. A variety of scholars conceptualized engagement based on: (a) the conditional context, such as institutions that promote social welfare, or classrooms’ learning activities (Skinner & Pitzer, 2012); or (b) modalities, such as momentary or long-term engagements (Sun & Rueda, 2011). There are at least two critical characteristics of learner engagement that are consistent among researchers with different foci despite using different definitions and categorizations to describe it. These characteristics include the fact that it is an action-driven process (Fredricks et al., 2016; Sulis & Philp, 2020), which means that engagement is characterized by 'energized, focused, and sustained actions' (Skinner et al., 2008, p. 225); as well as the fact that it is multidimensional and even overlaps among those dimensions.
The last few years have seen a boom in research on L2 learners' engagement (Hiver et al., 2020). The following three dimensions of learner engagement have been identified as generally relevant to foreign language contexts (Fredricks et al., 2016; Sulis & Philp, 2020). These have been conceptualized by the researchers based on the study context:
Behavioral engagement (BE): when learners pay attention to learning itself, asking questions and participating in exchanging information to generate initiatives.
Cognitive engagement (CE): the cognitive effort made by language learners to acquire the skills and knowledge required to accomplish complex tasks.
Emotional engagement (EE): students’ affective responses to interactions with teachers, peers, and the environment.
Students’ approaches to virtual exchange (SAVE)
Research on SAL dates back to Marton and Säljö (1976), who discovered that students process information at two levels: deep processing and surface processing. Entwistle (2018) eventually coined the term 'approaches to learning'. He defined the surface approach as a learning strategy that involves memorizing and reproducing learning material with the intention of simply learning facts to pass a course. Additionally, it has been associated with students' failure to recognize connections between ideas or concepts. The deep approach refers to a student's objective to gain a deeper understanding of the topic of study by utilizing critical thinking skills and meaningful learning to develop a personal understanding (Entwistle., 2018).
As technology has advanced, SAL has been validated, revised, and expanded by studies in blended learning (Ellis & Bliuc, 2015), online learning (Ellis & Bliuc, 2017; Han & Geng, 2023) and language Massive Open Online Course (LMOOCs, Rahimi, 2024). Researchers in the field have yet to apply SAL's application, validation, and definition to VEs. Based on this, we define the deep approach as a means of stimulating new modes of thinking through VE, which is closely related to students' intentions to interact with one another and share information, as well as their motivation to do so. A surface approach involves reducing interactions and meaningful online presence, and limiting VE usage.
The stimulus organism response (S-O-R) framework
A S-O-R framework describes the interaction between stimulus and response based on principles of environmental psychology (Mehrabian & Russell, 1974). Stimulus refers to factors that stimulate or facilitate action that influence individual states. Organisms mediate the effects of stimuli on responses in affective and cognitive states. Our study is well suited to the S-O-R model, as it illustrates the mechanisms and processes by which the type of VE task can reinforce learners' inner states since our goal is to examine the mechanisms by which ICAPCALL shapes learners' organisms (online engagement) and responses (SAVE). By applying this framework, we can provide another conceptual framework to the complex dynamic system of language learners to second language acquisition (SLA) since language learners' cognitive, affective, and behavioral domains are interconnected and their behavior changes in any SLA context (Freeborn et al., 2022; Rahimi, 2023; Rahimi & Cheraghi, 2022; Rahimi & Mosalli, 2024) that has not been applied to the VE context as far as we are aware, since most of VE studies in the field of CALL explored language learners learning behavior mainly through the qualitative research designs (Colpaert, 2020; Luo & Yang, 2018; O'Dowd, 2021), which failed to explore the serial correlations among the psycholinguistic factors in VE. Additionally, Colpaert (2020) and O'Dowd (2021) recommend that VE researchers emphasize the process-oriented side of VE alongside the outcome. Accordingly, we applied it to cover both process and product in our study.
Hypothesizes development
Studies focusing on language and cultural exchange have shown that the mode of interaction affects language learners' cultural competence. According to Yang (2017), synchronous interaction within VE influences language learners' critical thinking and reflection. Similarly, Luo and Gao (2022) found that synchronous and asynchronous VE positively impacted language learners' skills and cultural competence. However, some participants showed poor attitudes towards asynchronous VE because it was ineffective. Synchronous and asynchronous communication levels have been cited in relation to developing language learners' micro and macro skills, with asynchronous communication levels contributing significantly to pronunciation, cultural awareness, and self-esteem. Aside from VE's mode of interaction, previous studies have found its frequency crucial to shaping different skills and competencies. A study by Sevilla-Pavón and Nicolaou (2020) showed that learners were better equipped to address workplace demands by regularly integrating VE into university curricula. Additionally, previous studies have reported its frequency of use to improve learners' problem-solving skills (Ou Yang et al., 2023), motivation (Demir & Kayaoğlu, 2021), and positive attitudes toward it (Luo & Yang, 2021).
CALL researchers have explored how VE implementation frequencies and modes of interaction can influence ICC, motivation, and language learners' micro and macro skills. Our knowledge of CALL research indicates that VE has yet to be explored from the perspective of shaping language learners' engagement, particularly tri-phenomenon ones. As mentioned above, term engagement reflects the quality of learners' effort put forth in order to perform well and meet their objectives. Recent studies have highlighted that digital literacy (Zhang & Hyland, 2023), task complexity (Qiu, 2022), and the environment shape learners' engagement. However, a more pressing need is to determine whether the quality and type of tasks in VE contribute to learners' engagement, and their approaches to VE, accordingly the researchers developed the following hypothesize:
H1
Passive activities in VE will positively shape learners’ (H1a) emotional, (H1b) behavioral, and (H1c) cognitive engagements.
H2
Active activities in VE will positively shape learners’ (H2a) emotional, (H2b) behavioral, and (H2c) cognitive engagements.
H3
Constructive activities in VE will positively shape learners’ (H3a) emotional, (H3b) behavioral, and (H3c) cognitive engagements.
H4
Interactive activities in VE will positively shape learners’ (H4a) emotional, (H4b) behavioral, and (H4c) cognitive engagements.
H5
Passive activities in VE will positively shape learners’ (H5a) deep, and (H5b) surface approach.
H6
Active activities in VE will positively shape learners’ (H6a) deep, and (H6b) surface approach.
H7
Constructive activities in VE will positively shape learners’ (H7a) deep, and (H7b) surface approach.
H8
Interactive activities in VE will positively shape learners’ (H8a) deep, and (H8b) surface approach.
The role of learners' perceived support, emotion, and self-efficacy in shaping online learning approaches was investigated by Han and Geng (2023). They reported that instructional, peer, and technical support positively shaped learners' deep approach to online learning. Also, self-efficacy and emotion mediate these relationships. A mediation analysis also highlighted motivation's role in influencing university learners' approaches to target educational contexts (Schmidt, 2020) as well as LMOOC (Rahimi, 2024). This was also reported by Rahimi and Sevilla-Pavón (2024b), where online self-regulation has a crucial role as a mediator and moderator variable between their perceived instructional support and their learning approaches to online language learning. What those studies have neglected is the mediation role of engagement as one of the psychological factors involved. It is essential to understand engagement because it measures students' commitment and effort in relation to the educational context in which they wish to participate.
Students' approaches to learning have recently attracted scholars' attention as a means to understand how learners shaped their approaches to target educational settings and what factors influence those approaches. As an instance, Ellis et al. (2016) in their studies aimed to evaluate the quality of learners’ experienced in blended learning, online, and face-to faced classrooms. The cluster analysis showed that learners’ had deep approach to face to face than two others learning approaches. In another study Han and Geng (2023) investigated the role of learners’ perceived supports, emotion, and self-efficacy in shaping learners’ approach to online education. According to them the instructional, peer, and technical support positively shaped their deep approach to online learning, Moreover, both self-efficacy, and emotion mediated these relationships as well. In the recent studies, the mediation analysis underlined the role of motivation in shaping university learners’ approaches to target educational context (Zakariya et al., 2020). In the context of LMOOC, Rahimi (2024) reported that language learners' deep approach to LMOOC was directly predicted by instructional and peer support, which was mediated by learners' online motivational-self systems. As a result of these reports, learners' psychological factors significantly mediated the correlation between instructional, technical, and peer support, as well as their approach to the digital learning environment. Based on the previous findings and the literature gaps, we developed the following hypotheses to investigate another psychological factor that might act as predictor as well as mediator to shape learners' approaches to VE:
H9
The cognitive engagement will positively predict learners' deep approach and approach to the virtual exchange.
H10
The behavioral engagement will predict learners' deep approach and approach to the virtual exchange.
H11
The emotional engagement the behavioral engagement will predict learners' deep approach and approach to the virtual exchange.
H12
The cognitive engagement will positively mediate the quality integration of the virtual exchange and the learners' deep approach to the virtual exchange and negatively mediate the surface approach to the virtual exchange.
H13
The behavioral engagement will positively mediate the quality integration of the virtual exchange and the learners' deep approach to the virtual exchange and negatively mediate the surface approach to the virtual exchange.
H14
The emotional engagement will positively mediate the quality integration of the virtual exchange and the learners' deep approach to the virtual exchange and negatively mediate the surface approach to the virtual exchange.
Literature gap
A review of recent literature identified several gaps in the VE field. Our literature review revealed that VE usage and interaction modes influence language learners' ICCs, language skills, motivations, and digital competence (Luo & Yang, 2021; Sevilla-Pavón & Nicolaou, 2020). However, it is unclear how the quality of the VE task could affect learners' psychological behavior and performance. In addition, although recent studies have found that VE influences learners' motivation or perceptions (Payant & Zuniga, 2022) about it, how it influences learners' online engagement and approaches remains unclear. Even though students' approaches to the educational context are already discussed about traditional (Coertjens, 2018), online (Han & Geng, 2023) LMOOC (Rahimi, 2024), and blended education (Niu et al., 2022), VE remains overlooked. As far as we know, no prior study in the CALL field applied the exploratory approach to develop a VE conceptual model. Hence, using the S-O-R framework, we will construct a conceptual framework tailored to VE in CALL. A conceptual framework will be formulated by ICAPCALL, which might facilitate learners' online engagement and act as a stimulus for an organism. Our conceptual model also includes SAVE as a response in which the organism mediates stimulus influence. Our conceptual framework is shown in Fig. 1.
[See PDF for image]
Fig. 1
The study hypothesize model
Methodology
Study context and procedure
92 students majoring in English Studies at the Universidad de Valencia 1 in Spain, in Business and Spanish at the Cyprus university of technology 2, and in Public Relations at Limerick university participated in the study. During the first semester of the 2023–24 academic year, the Spanish cohort exchanged cultural and historical information with Irish students in English and Spanish and Cypriot English students. Most participants were females (74), followed by males (18). 73 participants were between 18 and 20, 15 between 21 and 24, and 4 over 24.46 participants had studied the target language for more than six years, while three studied it for one to three years. Table 1 summarizes demographic information.
Table 1. Participants’ demographic information
N | % | ||
|---|---|---|---|
Gender | Male | 18 | 18.9 |
Female | 72 | 80.1 | |
Years | |||
Age | 18–20 | 73 | 83 |
21–24 | 15 | 16 | |
24 < | 4 | 1 | |
Years | |||
Language | 1–3 | 3 | 6.3 |
Learning experience | 4–5 | 46 | 49.5 |
6 < | 43 | 44.2 |
From 2021 to the end of 2024, this study was conducted as part of a national project funded by Spanish Ministry of Science and Innovation. The VE tasks were completed in translational groups of four to six students. In our project titled developing Socioemotional Skills Through Virtual Exchange (SOCIEMOVE), artificial intelligence (AI) was combined with immersive, three-dimensional virtual exchanges (3DVE) on the virtual reality platform Spatial. There were three synchronous online meetings during the Fall semester of 2023. The project also involved weekly asynchronous communication, such as Google Classroom and WhatsApp. On Moodle, teachers provided learning content based on their course syllabus, and define some cultural elements for learners (passive). Then on Google Classrooms, participants exchanged self-introduction videos and ice-breaking questions to break the ice. Next, they used the Hello History AI app to interview a historical figure from the target culture (e.g., from an English-speaking country in the case of Spanish and Cypriot students and from a Hispanic country in the case of Irish students). Afterwards, they used the AI interview and additional sources, including books and online resources, to share the information they had gathered about their respective historical figures (active and constructive learning). Information was shared on the Spatial VR platform through an immersive, 3D synchronous meeting (constructive learning). Following that, students worked collaboratively on their digital stories, which were based on their chosen historical figures (Interactive learning) and constituted the main final output of the VE initiative. For each set of tasks, students were required to write a reflective essay (constructive and interactive learning), which was then shared in their groups for peer feedback. Afterwards, the essays were uploaded to their university's Moodle learning management system. The aforementioned artifact was then collaboratively co-constructed in different international groups as a digital story, combining different media types. As part of this process, the students wrote a script, created a storyboard for their digital stories, and then shared it with their peers for feedback and suggestions (interactive, and constructive). In the subsequent versions, students wrote the script until it was ready for voice-over. This was then coordinated with a video sequence, an original soundtrack, subtitles in different languages, and the digital story's final edition (active learning and constructive learning). When the digital stories were exported to a suitable format and posted to the Google Classroom Community, participants were asked to discuss the digital stories like a film review (interactive learning). They then used a peer assessment rubric to evaluate the digital stories and voted on the best. In an award-giving ceremony organized by students, the most voted digital stories were exhibited and awarded.
Instruments
The study instrument including its items, sources, and normality distribution are shown as an appendix. The items were rated from 1 = strongly disagree to 5 = strongly agree. Additionally, the data were normally distributed between − 1 and 1.
Data analysis
We used a bi-symmetrical research design that was first introduced and used in education, applied linguistics, and CALL by Rahimi and Sevilla-Pavón (2024a), where the symmetrical approach is PLS-SEM as its ordinary least squares and principal component analysis, specifying it to systematically develop a conceptual model. It follows additive sufficiency logic to identify the should-have factors that contribute to predict the target outcome (Rahimi & Sevilla-Pavón, 2024a). Furthermore, PLS-SEM is a statistical method focusing on theory development in light of its predictive nature (Hair et al., 2021). As Hair et al. (2020) highlighted, PLS-SEM can estimate direct, indirect, and serial interactions between latent variables simultaneously, which is consistent with the S-O-R model that we propose.
Additionally, recent CALL studies have extensively applied qualitative design (O’Dowd & Dooly, 2021), mixed methods (Gruber et al., 2023), and action research (Dooly & Sadler, 2019) in VE. However, CALL researchers have not developed any theories in this context so far, or apply bi-symmetrical approach.
In the asymmetric phase of our data analysis, NCA was used, which was developed by Dull (2016). It is important to integrate these two quantitative approaches because the symmetrical approach uses the average mean scores of indigenous variables over exogenous ones to evaluate the prediction (fuzzy logic), whereas NCA follows identify the must-have factors that predict the target outcome (necessary logic; Rahimi & Sevilla-Pavón, 2024a). Accordingly, integrating both approaches can generate more results to the study.
Results
Symmetrical analysis
The formative phase
The PLS-SEM reflective phase involves calculating latent variables' validity and reliability. Cronbach Alpha and Composite Reliability were used to determine reliability. Validity was also determined using convergent and discriminant validity. Hair et al. (2021) suggest that reliability should be cut to 0.7 and validity at 0.5. Table 2 shows high reliability and validity for all variables.
Table 2. Study constructs' reliability and validity
Variables | Cronbach's alpha | Composite reliability (rho_a) | Composite reliability (rho_c) | Average variance extracted (AVE) |
|---|---|---|---|---|
AL | 0.877 | 0.878 | 0.924 | 0.803 |
BE | 0.814 | 0.828 | 0.889 | 0.729 |
CE | 0.853 | 0.853 | 0.911 | 0.772 |
CL | 0.891 | 0.897 | 0.932 | 0.821 |
DA | 0.831 | 0.839 | 0.898 | 0.747 |
EE | 0.859 | 0.864 | 0.914 | 0.780 |
IL | 0.882 | 0.886 | 0.927 | 0.809 |
PL | 0.806 | 0.811 | 0.886 | 0.721 |
SA | 0.824 | 0.824 | 0.895 | 0.740 |
To evaluate discriminant validity, we applied the Heterotrait-monotrait ratio (HTMT)–Matrix which was recently developed by Henseler et al. (2015). It evaluates the correlation between indicator items and other indicators' mean values. It should be less than 0.85. Table 3 shows that all variables had discriminant validity.
Table 3. Discriminant validity
AL | BE | CE | CL | DA | EE | IL | PL | SA | |
|---|---|---|---|---|---|---|---|---|---|
AL | |||||||||
BE | 0.280 | ||||||||
CE | 0.485 | 0.712 | |||||||
CL | 0.405 | 0.495 | 0.622 | ||||||
DA | 0.389 | 0.526 | 0.753 | 0.579 | |||||
EE | 0.319 | 0.773 | 0.616 | 0.704 | 0.507 | ||||
IL | 0.298 | 0.581 | 0.589 | 0.542 | 0.544 | 0.741 | |||
PL | 0.193 | 0.135 | 0.209 | 0.057 | 0.278 | 0.246 | 0.174 | ||
SA | 0.196 | 0.659 | 0.779 | 0.472 | 0.549 | 0.490 | 0.437 | 0.124 |
Formative model
In the formative model evaluation of our symmetrical phase we incorporated a variety of metrics to generate our conceptual model and test our hypotheses, including multicollinearity, path coefficient, t-value, coefficient of determination (R2), assessing the direct, and indirect correlations, Stone-Geisser value (Geisser, 1974; Stone, 1976), Standardized Root Mean Square Residual SRMRs, and Goodness of Fit indices (GOF).
The variance inflation factor (VIF) was used to measure multicollinearity. According to Hair et al. (2021), endogenous variables should have a VIF index lower than 5, which our study had. As a next metric we assessed our hypotheses based on their path coefficient, t-values, and p-value. By applying 5000 bootstrap sample analysis we found that AL- > CE (), IL- > EE, (), IL- > BE (), IL- > CE (), CL- > EE (), CL- > BE (), CL- > CE, CE- > DA (), CE- > SA (), and BE- > SA () had significant correlation. The rest were non-significant because their t-values were lower than 1.96, or their p-values were higher than 0.05. Table 4 summarizes the bootstrap results.
Table 4. Result of the bootstrap analysis
Research hypothesis | Path exogenous -------- > endogenous | β | t-value | P value | VIF |
|---|---|---|---|---|---|
H1a | PL- > EE | 0.127 | 1.557 | 0.119 | 1.043 |
H1b | PL- > BE | 0.298 | 0.034 | 0.765 | 1.043 |
H1c | PL- > CE | 0.864 | 0.034 | 0.387 | 1.043 |
H2a | AL- > EE | − 0.006 | 0.066 | 0.947 | 1.86 |
H2b | AL- > BE | 0.049 | 0.420 | 0.675 | 1.86 |
H2c | AL- > CE | 0.216 | 2.293 | 0.022 | 1.86 |
H3a | IL- > EE | 0.436 | 5.760 | 0.000 | 1.345 |
H3b | IL- > BE | 0.359 | 3.544 | 0.000 | 1.345 |
H3c | IL- > CE | 0.288 | 3.112 | 0.002 | 1.345 |
H4a | CL- > EE | 0.403 | 4.032 | 0.000 | 1.423 |
H4b | CL- > BE | 0.240 | 2.375 | 0.018 | 1.423 |
H4c | CL- > CE | 0.323 | 3.134 | 0.002 | 1.423 |
H9a | CE- > DA | 0.552 | 5.646 | 0.000 | 1.825 |
H9b | CE- > SA | − 0.513 | 5.901 | 0.000 | 1.825 |
H10a | BE- > DA | 0.033 | 0.318 | 0.751 | 2.053 |
H10b | BE- > SA | − 0.248 | 2.703 | 0.007 | 2.053 |
H11a | EE- > DA | 0.123 | 1.245 | 0.213 | 1.825 |
H11b | EE- > SA | 0.017 | 0.207 | 0.836 | 1.825 |
The direct bootstrap analysis with high confident intervals (97.5% CI) showed that constructive learning and the interactive one significantly shaped learners’ deep (CL- > DA; ); (IL- > DA) and negatively shaped their surface approach (CL- > SA; ); (IL- > SA = − ). Table 5 presents a direct bootstrap.
Table 5. Direct bootstrap analysis
Hypothesizes | Path | β | 2.5% | 97.5% | t value | p value |
|---|---|---|---|---|---|---|
H5a | AL- > DA | 0.120 | 0.009 | 0.260 | 1.877 | 0.061 |
H5b | AL- > SA | − 0.123 | − 0.255 | 0.016 | 1.768 | 0.077 |
H6a | CL- > DA | 0.236 | 0.103 | 0.362 | 3.632 | 0.000 |
H6b | CL- > SA | − 0.218 | − 0.369 | − 0.080 | 2.957 | 0.003 |
H7a | IL- > DA | 0.224 | 0.102 | 0.340 | 3.713 | 0.000 |
H7b | IL- > SA | − 0.230 | − 0.360 | − 0.081 | 3.226 | 0.001 |
H8a | PL- > DA | 0.061 | − 0.052 | 0.179 | 1.062 | 0.288 |
H8b | PL- > SA | − 0.048 | − 0.188 | 0.082 | 0.690 | 0.490 |
Furthermore, the mediation analysis showed that there were seven serial correlations positively shaping learners’ deep (AL- > CE- > DA; 0.119); (IL- > CE- > DA; ), (CL- > CE- > DA; ) and negatively surface approach (IL- > BE- > SA; − 0.089;) (IL- > CE- > SA; ), (CL- > CE- > SA; ), (AL- > CE- > SA; ). Table 6 displays the result of the mediation analysis.
Table 6. Result of the serial mediation analysis
Hypothesizes | Path | β | 2.5% | 97.5% | t value | p value |
|---|---|---|---|---|---|---|
H12 | CL- > CE- > SA | − 0.165 | − 0.306 | − 0.051 | 2.549 | 0.011 |
PL- > CE- > SA | − 0.042 | − 0.145 | 0.050 | 0.837 | 0.403 | |
IL- > CE- > SA | − 0.148 | − 0.249 | − 0.047 | 2.876 | 0.004 | |
AL- > CE- > SA | − 0.111 | − 0.220 | − 0.019 | 2.151 | 0.031 | |
AL- > CE- > DA | 0.119 | 0.020 | 0.238 | 2.134 | 0.033 | |
IL- > CE- > DA | 0.159 | 0.046 | 0.291 | 2.541 | 0.011 | |
CL- > CE- > DA | 0.178 | 0.057 | 0.314 | 2.735 | 0.006 | |
PL- > CE- > DA | 0.045 | − 0.055 | 0.160 | 0.824 | 0.410 | |
H13 | CL- > BE- > SA | − 0.059 | − 0.156 | − 0.006 | 1.506 | 0.132 |
PL- > BE- > SA | − 0.009 | − 0.083 | 0.046 | 0.269 | 0.788 | |
IL- > BE- > SA | − 0.089 | − 0.183 | − 0.022 | 2.102 | 0.036 | |
AL- > BE- > SA | − 0.012 | − 0.071 | 0.059 | 0.394 | 0.694 | |
CL- > BE- > DA | 0.008 | − 0.054 | 0.059 | 0.288 | 0.773 | |
PL- > BE- > DA | 0.001 | − 0.028 | 0.030 | 0.087 | 0.931 | |
IL- > BE- > DA | 0.012 | − 0.076 | 0.083 | 0.303 | 0.762 | |
AL- > BE- > DA | 0.002 | − 0.020 | 0.039 | 0.115 | 0.908 | |
H14 | CL- > EE- > SA | 0.007 | − 0.055 | 0.073 | 0.208 | 0.835 |
PL- > EE- > SA | 0.002 | − 0.021 | 0.032 | 0.167 | 0.867 | |
IL- > EE- > SA | 0.007 | − 0.059 | 0.085 | 0.201 | 0.841 | |
AL- > EE- > SA | 0.000 | − 0.013 | 0.020 | 0.013 | 0.990 | |
CL- > EE- > DA | 0.049 | − 0.030 | 0.130 | 1.236 | 0.217 | |
PL- > EE- > DA | 0.016 | − 0.008 | 0.070 | 0.771 | 0.441 | |
IL- > EE- > DA | 0.053 | − 0.032 | 0.147 | 1.196 | 0.232 | |
AL- > EE- > DA | − 0.001 | − 0.027 | 0.037 | 0.050 | 0.960 |
Moreover, the coefficient of determination (R2) was applied to explore the predictive power of the exogenous variables on the indigenous ones. Hair et al. (2021) reported weak, moderate, and strong predictive power for the model at 0.19, 0.37, and 0.067 as shown in Table 7. Accordingly, BE (0.29), CE (0.42), DA (0.42), EE (0.55) and SA (0.46) were obtained at a sufficient level. Also Stone-Geisser's (Q2) value was used to evaluate model accuracy, and it exceeded zero. The model fit indices were evaluated simultaneously using two metrics. Accordingly, the SRMR achieved below 0.08 and the GOF reached 0.61, which indicates a well-fitted model. The study final model presents in Fig. 2.
Table 7. Results of the structural model
Variables | ||
|---|---|---|
BE | 0.191 | 0.296 |
CE | 0.306 | 0.428 |
DA | 0.292 | 0.420 |
EE | 0.413 | 0.556 |
SA | 0.323 | 0.463 |
SRMR = 0.06 | ||
[See PDF for image]
Fig. 2
Study model with relative value
Necessary conditional analysis (NCA)
Asymmetrical NCA was applied to our study, as discussed earlier. Therefore, the latent variable score was derived from the symmetrical part to the current part. In contrast to PLS-SEM, NCA is sensitive to normality distributions. As with linear regression, it is limited to one independent variable. Thus, we explored the role of ICAPCALL and online engagement in shaping learners' deep approach to VE. As shown in Table 8, we calculated the normality distribution of the latent variables' scores and they were normally distributed between − 2 and 2.
Table 8. Latent variable scores' normality distribution
Latent variables’ score | Mean | Observed min | Observed max | Excess kurtosis | Skewness |
|---|---|---|---|---|---|
CL | 0.000 | − 1.939 | 2.195 | − 1.027 | 0.144 |
EE | 0.000 | − 2.900 | 1.875 | − 0.181 | − 0.417 |
DA | 0.000 | − 2.699 | 2.160 | 0.026 | − 0.445 |
BE | 0.000 | − 2.735 | 2.140 | − 0.238 | − 0.188 |
IL | 0.000 | − 2.225 | 1.771 | − 0.688 | − 0.053 |
PL | 0.000 | − 2.520 | 2.447 | − 0.225 | − 0.327 |
EG | 0.000 | − 2.446 | 2.044 | − 0.896 | − 0.299 |
AL | 0.000 | − 2.322 | 2.052 | − 0.519 | − 0.290 |
To report our NCA result we applied ceiling envelopment–free disposal hull (CE-FDH), since its accuracy per definition is 100% while ceiling regression–free disposal hull (CR-FDH) is less than this. According to ceiling line effect size, all variables except emotional engagement (p > 0.05) contributed to shaping learners’ deep approach to VE. Dull (2016) further notes that d values below 0.1 are considered small, those between 0.1 and 0.3 are considered median, those between 0.3 and 0.5 are considered large, and those above 0.5 are considered very large. According to Table 9, all variables had a median impact.
Table 9. The result of the NCA
Variables | CE-FDH | p value |
|---|---|---|
AL | 0.158 | 0.044 |
BE | 0.205 | 0.034 |
CL | 0.148 | 0.037 |
EE | 0.194 | 0.124 |
CE | 0.275 | 0.000 |
IL | 0.168 | 0.044 |
PL | 0.164 | 0.038 |
Furthermore, according to the bottleneck, there were eight conditions to shape learners’ deep approach to virtual exchange. For clarification, in order to achieve 50% level of deep approach, five conditions need to be in place. For this sake, active learning should be no less than 32%, and behavioral engagement no less than 21%, while cognitive engagement, interactive, and passive learning should be no less than 21%. The following conditions are shown in Table 10. Figures 3, 4, 5, 6, 7, 8 and 9 display the NCA charts for each variable.
Table 10. The bottleneck of deep approach
Deep | AL | BE | CL | CE | IL | PL | |
|---|---|---|---|---|---|---|---|
0.000% | − 2.699 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
10.000% | − 2.213 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
20.000% | − 1.727 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
30.000% | − 1.241 | 0.000 | 0.000 | 0.000 | 2.174 | 0.000 | 0.000 |
40.000% | − 0.756 | 0.000 | 0.000 | 0.000 | 2.174 | 0.000 | 0.000 |
50.000% | − 0.270 | 3.261 | 2.174 | 0.000 | 2.174 | 2.174 | 2.174 |
60.000% | 0.216 | 3.261 | 2.174 | 0.000 | 2.174 | 2.174 | 2.174 |
70.000% | 0.702 | 3.261 | 15.217 | 18.478 | 8.696 | 2.174 | 15.217 |
80.000% | 1.188 | 29.348 | 15.217 | 18.478 | 31.522 | 26.087 | 15.217 |
90.000% | 1.674 | 29.348 | 15.217 | 18.478 | 66.304 | 26.087 | 15.217 |
100.000% | 2.160 | 29.348 | 86.957 | 88.043 | 88.043 | 78.261 | 73.913 |
[See PDF for image]
Fig. 3
The NCA of active learning
[See PDF for image]
Fig. 4
The NCA of passive learning
[See PDF for image]
Fig. 5
The NCA of constructive learning
[See PDF for image]
Fig. 6
The NCA of interactive learning
[See PDF for image]
Fig. 7
The NCA of emotional engagement
[See PDF for image]
Fig. 8
The NCA of behavioral engagement
[See PDF for image]
Fig. 9
The NCA of cognitive engagement
Several scenarios were identified by combining the symmetric and asymmetric results. First, some variables, such as active and passive learning, are necessary to shape learners' deep understanding of VE; however, they didn't manifest it further. Additionally, some variables, such as constructive learning, interactive learning, behavioral engagement, and cognitive engagement, were necessary for learners to develop deep approaches to VE, and additional levels of these variables will facilitate this development. Due to its non-significance, emotional engagement wasn't necessary, nor did it improve learners' deep approach to VE. Table 11 presents the relevant scenarios for interpreting the study's findings.
Table 11. Relevant scenarios to interpret the findings
Variable | Symmetric phase | Asymmetric phase | Conclusion |
|---|---|---|---|
Active learning | Not significant | Necessary | A certain level of active learning was necessary to shape learners’ deep approach. However, a further level of it did not escalate learners’ deep approach. |
Passive learning | Not significant | Necessary | A certain level of passive learning was necessary to shape learners’ deep approach. However, a further level of it did not increase learners’ deep approach. |
Constructive learning | Significant | Necessary | A certain level of constructive learning was necessary to manifest learners’ deep approach. Moreover, a further level of it improved their deep approach. |
Interactive learning | Significant | Necessary | A certain level of interactive learning was necessary to manifest learners’ deep approach. In addition, a further level of it increased students’ deep approach. |
Behavioral engagement | Significant | Necessary | To manifest learners' deep approach, a certain level of behavioral engagement is required. A further level of it also increased it. |
Emotional engagement | Not significant | Not necessary | Non-significant |
Cognitive engagement | Significant | Necessary | A certain degree of cognitive engagement is required to manifest learners' deep approach. Further levels of it also enhanced it. |
Discussion
The factorial structure of ICAPCALL, SAVE and online engagement in VE
Based on the PLS-SEM formative phase, we validated the ICAPCALL, SAVE and online learning engagement in the Spanish EFL and VE contexts. These findings complement previous research that validated learners' approaches to educational contexts in a variety of educational settings, including online (Han & Geng, 2023), online language learning (Rahimi & Sevilla-Pavón, 2024b), LMOOC (Rahimi, 2024) and blended (Niu et al., 2022) learning contexts. We also validated a special version of the ICAP for the CALL field and language learners, since the original ICAP was only intended for educators (Han & Geng, 2023). More specifically, O’Dowd and Ware (2009) classified VE tasks in three types: information exchange, comparison and analysis, and collaborative tasks. Alongside these classifications we added to the CALL literature and validated four types of tasks in VE, including passive, active, interactive, and constructive. We also validated the three-dimensional aspects of online engagement in VE, as they had previously been overlooked by VE researchers, but have been validated in recent studies for Online language learning (Sun & Zhang, 2024), and online courses (Gunness et al., 2023).
The bisymmetric results
As a result of the PLS-SEM reflective phase, passive learning did not significantly influence learners’ engagement and their approaches to VE. However, the asymmetrical part of the study revealed that it was among the necessary conditions for learners to develop their deep approach to virtual exchange. Nevertheless, a higher level of it might not manifest the deeper approach of learners. This aligns with “knowledge-change processes” (Chi, 2009), since passive learning is the beginning of the dynamic process of learners’ engagement to become active, interactive, and constructive learners, as they significantly shaped learners’ engagements and deep approach.
Furthermore, active learning only predicted learners’ cognitive engagement (β = 0.216) and failed to predicted other types. Thus, the greater the students' opportunities to apply new knowledge, the greater their cognitive efforts will be dedicated to achieving their learning objectives in VE. This might come from our previous finding, which showed that passive learning might not generate learners’ engagement but rather it generates more cognitive efforts, which is the case for (AL- > CE). It was also found among the necessary conditions to shape learners’ deep approach but it did not elevate it. This finding is in accordance with Gruber et al.'s (2023), who reported that the levels of learners’ activeness in VE depended on how they perceived its real-life connection with the target context.
It was also observed that interactive learning activities positively influenced learners' engagement, illustrating that the more cooperative language learners were in their VE activities, and the more they built their knowledge based on their background information and interaction with others, the more emotionally, cognitively, and behaviorally engaged they were. Thus, when students co-construct knowledge based on information exchange, their engagement in activities might be higher. Moreover, an interesting finding among these correlations is that interactive learning tasks had a higher shared variance with learners' emotional engagement (β = 0.436), suggesting the interactive activities in VE elicit stronger affective responses from the learners, as they interact more with teachers and peers within the VE to construct their own knowledge. This finding added an element of engagement to the literature in addition to contributing to the open-interactional aspects of VE that previous CALL scholars have already reported as having a significant impact on learners' ICC (Hafour et al., 2023), language skills (Demir & Kayaoğlu, 2021) and 21st-century digital competence (Sevilla-Pavón & Nicolaou, 2020). Conversely, the asymmetrical analysis underscored that interactive learning tasks were necessary to shape learners' deep approach. Unlike the two previous activities, it directly contributed to escalating it as well, which might be due to the fact that learners had to reach an agreement on their project results (Luo & Yang, 2018), culminating in developing new modes of thinking, and continuing to exchange ideas with others, since to have good interactive tasks, learners’ should be constructive (Chi & Wylie, 2014).
Constructive learning emerges in learning tasks in which learners individually reach the learning outcomes based on their own reflection. According to the results, it significantly generated organisms, which are comprised of learners' three-dimensional engagements. In this way, constructive learning that requires learners to achieve the VE outcomes on their own has resulted in learners asking their peers questions and acquiring new skills and knowledge to engage in more initiative-generating learning activities. This finding leads us to draw a different conclusion from another perspective, since VE figureheads and scholars often tend to believe that VE's collaborative nature is the source of its success (Demir & Kayaoğlu, 2021; O’Dowd, 2021). Based on these correlations (CL- > EE; CL- > BE; CL- > EE), along with the collaborative nature of VE, engagement and a deep approach to learning from learners themselves, building knowledge and collaborating with others, played an important role in its success. Accordingly, alongside the frequency use of VE (Sevilla-Pavón & Nicolaou, 2020), the mode of interaction (Luo & Gao, 2022; Yang, 2017) and its collaborative nature (Demir & Kayaoğlu, 2021), its quality integration and learners’ individual skills were also among the most critical success factors, which were dependent upon how VE was integrated. In fact, we should not only focus on the collaborative nature of VE as a successful factor in CALL, since other CALL initiatives such as LMOOCs have the same collaboration opportunity as well, but have nevertheless a low completion rate. Thus, there are other individual, psychological and contextual factors involved in any success of CALL materials which should be considered.
Considering the S-O-R model, all stimulus components except passive learning significantly impacted learners' organisms. Accordingly, there is a need to further investigate the role of organisms in facilitating the response. In this regard, the structural model showed that learners' cognitive engagement positively impacted their deep (β = 0.552) and negatively impacted their surface approach (β = − 0.513). A possible explanation may be that this finding has to do with learning tasks (stimuli), since learning activities that require learners’ involvement in thinking, collaborating, solving problems, making decisions and selecting or integrating knowledge may cognitively activate them to be more engaged in the learning context (Fütterer et al., 2023; Rahimi, 2024). It should be noted that, of all the organism components, cognitive engagement had the highest positive and negative shared variance with both surface and deep approaches. Possibly, this result might depend on the interest-to-effort ratio (Rahimi, 2024; Smarandache et al., 2021), with uninteresting tasks being associated with surface approaches and minimal online presence, and interesting tasks being supported by learners and meaningful online presence. The results of our serial mediation also support this conclusion, as cognitive engagement positively mediated three types of tasks: active, interactive and constructive. Each has been shown to have different procedures, culminating in VE being used effectively by learners to stimulate new ways of thinking (the deep approach).
Additionally, behavioral engagement only influenced learners' surface approach to VE, while it had no impact on learners' deep approach. Having given attention to learning itself, exchanging more information, and participating in VE, students will maximize their online presence in VE. It is possible that the supportive environment of VE might relate to this result, since recent surveys indicated that learners' sense of connectedness with others is influenced by their approach to learning on both a deep and a surface level (Forsblom et al., 2021; Han & Geng, 2023). A further finding was that emotional engagement was not associated with any approach taken by learners to VE. Based on this result, O’Dowd et al., (2019) were right when they claimed that simply engaging learners in VE is not sufficient to ensure intercultural learning success.
In our deep dive into the S-O-R and serial mediation analyses, we discovered that some organism components had a flipping effect on the direction between stimulus and response. Interestingly, only CL and IL positively predicted learners' deep (CL- > DA; ); (IL- > DA; ) and negatively predicted surface approaches to VE in the direct analysis (CL- > SA; ); (IL- > SA = − ); while in the mediation analysis (Table 6), cognitive and behavioral engagements significantly mediated between active tasks and learners' deep and surface approaches. However, emotional, as well as behavioral engagement did not seem to mediate learners' deep and surface approaches, and acted as moderator variables in some serial correlations. For clarification behavioral engagement change the significant direct correlations to insignificant ones in (CL- > BE- > SA; ; CL- > BE- > DA; ; IL- > BE- > DA; ; AL- > BE- > DA; ), and emotional engagement acted as moderator variables in (CL- > EE- > SA;; PL- > EE- > SA; ; AL- > EE- > SA; IL- > EE- > SA; ; CL- > EE- > DA; PL- > EE- > DA; IL- > EE- > DA; ; AL- > EE- > DA). This might be due to the learners' individual skills which is the case in constructive learning, which may be rooted in their preferences during their intercultural encounters and engagement with VE (Gruber et al., 2023). Furthermore, our results indicated that these preferences were determined by the way in which VE was applied based on ICAPCALL. Moreover, our results have expanded the literature by demonstrating that online engagement (Fredricks et al., 2016; Sulis & Philp, 2020), along with the levels of learners' emotion (Han & Geng, 2023), and motivation (Schmidt, 2020; Zakariya et al., 2020) that have been found, play a critical role in determining learners' approach to the target educational environment.
Recent studies have highlighted the collaborative nature of VE as one of the factors that have contributed to its success (Hafour et al., 2023), so we came to a new conclusion from a different point of view. As a result of this flipping effect, learners' cognitive engagement rather than emotional engagement (which reveals learners' collaborative engagement), primarily resulting from active, interactive, and constructive activities, was shown to have a major impact on their deep approach. Thus, it is the learners' cognitive effort that determines the success of VE alongside the social and collaborative aspects of it.
Conclusions
Having developed and validated our conceptual framework, the results of the symmetric phase of the study support the assertion that the current model is accurate in predicting 42% of language learners' deep, as well as 46% of surface approaches to VE. Furthermore, the asymmetric approach of our study showed that learners' deep approach to VE was shaped by six conditions. Based on these results, our study will have theoretical and practical implications in CALL.
Theoretical contribution
As a result of the study, there are several aspects that contribute to the literature in the fields of Applied Linguistics, Psycholinguistics, CALL, and VE, as follows:
The first step was to validate three conceptual frameworks for the literature: ICAPCALL, SAVE, and online engagement in the VE and CALL fields. In the second step, we moved the focus from the type of interactions in VE (Yang, 2017) and the frequency of its use (Sevilla-Pavón & Nicolaou, 2020) to the quality of its integration. Additionally, we moved further from the exploration of the role of VE in facilitating the development of learners' language skills and ICC (Luo & Gui, 2019) to their psychological aspects. Moreover, our findings expand the literature by demonstrating that online engagement is an important factor for students’ approach to a given educational environment, as recent findings only shed light on learners' motivation (Schmidt, 2020; Zakariya et al., 2020) and emotion (Han & Geng, 2023) as psychological factors. The interconnections between the online engagements of the learners and their deep and surface approaches to VE provided us with a new insight into the dynamic complex system (Freeborn et al., 2022; Rahimi, 2023; Rahimi & Cheraghi, 2022; Rahimi & Mosalli, 2024; Rahimi et al., 2025) of language learners in another SAL context.
Our study is also among the first, to the best of our knowledge, to apply multilevel analysis, including both direct, indirect and serial mediation, since recent studies have largely utilized case studies (Ware & Kessler, 2014), action research (Luo & Gui, 2019) or mixed methods (Chen et al., 2013). Additionally, we switched from integrating qualitative and quantitative analysis to integrating asymmetrical and symmetrical quantitative analysis simultaneously to generate results from both perspectives. To the best of our knowledge, this is also the first study to employ NCA in Applied Linguistics and CALL. The study is also among the first to follow the S-O-R framework when addressing the quality integration of VE (stimulus), shaping learners' online engagement (organism) which in turn facilitated learners' approaches (response). Our new conceptual framework for the study addressed both the process and product of VE, following the recommendations of the figureheads and systematic literature review which addressed these gaps (Colpaert, 2020; Luo & Yang, 2018; O’Dowd, 2021).
Practical implications
It is important that teachers do not solely focus on collaborating and engaging with the various learning groups across multiple contexts in VE but instead pay attention to learners' individual and cognitive engagement as well. As a matter of fact, teachers should include both collaborative and individual activities in their VE procedure. Aa an instance the personalization feature of artificial intelligence, which provides knowledge and feedback to students based on their context and level, can be used as a strategy for utilizing it in constructive and individual activities involving learners in hands-on activities (active tasks). Similarly, in our project that we used the Hello history app and asked learners to conduct interviews with it, and then to write scripts and create digital storytelling to demonstrate what they gained from these constructive, and individual learning activities through AI. Teachers can use any AI, and chatbots to provide constructive, and individual tasks.
In regards to collaborative and interactive learning activities, teachers can provide a collaborative environment using VR, Google Meet, and Zoom, and ask learners to share what they have already learned in their personalized learning task with artificial intelligence and ask them to provide feedback to each other to gain more knowledge, evaluate their works, and their activities that they have already done in their individual and constructive learning assignments.
In addition, VE projects with virtual reality environments are recommended for use in active and constructive activities as well as collaborative ones. Teachers can also use ICAPCALL to reflect on their VE integration or in any other CALL domain. Additionally, to shape learners' deep approaches, there were six conditions based on our NCA; teachers can follow each of these conditions based on their specific needs and according to their context.
When teaching in a context where resources are limited, teachers can use any CALL materials that are available to them to conduct hands-on activities, especially now that artificial intelligence large language models such as the Generative Pre-Trained Transformer (ChatGPT) are accessible to anyone. Teachers can take advantage of this accessibility, especially if they wish to follow ICAPCALL, and it can also be applied to the constructive and interactive phases for language learners in which they are able to individually develop concepts, compare them, solve their problems with the assistance of artificial intelligence, and then share and evaluate the results during the interactive phase.
At the next level, we recommend that teacher trainers and pedagogical experts train their teachers and raise their awareness about the possibilities brought about by the integration of VE in accordance with ICAPCALL. Thus, they should move beyond training them in CALL literacy, technological pedagogical content knowledge (TPACK), and digital literacy to how they can integrate VE, or any kind of ICT. As an evaluation tool, our ICAPCALL can also be used to evaluate the quality of ICT integration by in-service teachers or as a recruitment tool for pre-service teachers.
Additionally, pedagogical experts should encourage language teachers to develop high-quality objectives for the integration of VE and take a step further to elevate learners' language skills or intercultural competence with a view towards shaping their online engagement and approaches to this environment. In fact, no other outcome is more important than fostering a higher level of engagement and stimulating new modes of thinking within this context.
In macro terms, we developed and validated three conceptual frameworks for CALL and one model for VE. By doing so, we provide a new avenue for CALL researchers to apply, evaluate, extend and validate our framework in other CALL subfields, as well as in different EFL or ESL contexts. Moreover, the ICAPCALL model can also be used to measure digital transformation progress in CALL fields such as blended learning, online learning and flipped teaching and learning. Finally, the SAVE model can also be used as a measure to explore learners' success in any CALL subfields, particularly VE.
Limitations and future research
This study is merely a first step in exploring the role of ICAPCALL in shaping students' approaches to VE and their online engagement, particularly from the lens of the students. In light of this, ICAPCALL, SAVE, and online engagement can be applied, extended, and validated in other EFL and ESL contexts and CALL disciplines. Furthermore, we used self-reported data and questionnaire items to explore learners' online engagements and approaches to VE, which has a common limitation, and potential biases (e.g., social desirability or recall biases). The bias can be mitigated through the use of memory aids such as visual timelines, diaries, or other prompts that can assist participants in recalling events more accurately. Moreover, participants' actual behavior should be assessed, particularly for those variables that flipped the results in some directions (e.g., emotional engagement). To fill this gap, future studies need to apply longitudinal studies, case studies, or observation. Future studies should also apply ICAPCALL, SAVE, and learners' online engagement in other CALL fields. Furthermore, we recommend that future studies explore behavioral observations or longitudinal studies in order to validate the current findings and to apply the same procedures, theories, and tools in different learning environments.
Author contributions
Author 1: Theoretical and methodological framework, study conception and design, analysis and interpretation of results, draft manuscript preparation, reviewing the results and approving the final version of the manuscript. Author 2: Theoretical and methodological framework, proofreading, data collection, draft manuscript preparation, reviewing the results and approving the final version of the manuscript.
Funding
Acknowledgements are due to the Spanish Ministry of Science and Innovation for funding the research project SOCIEMOVE: DEVELOPING SOCIOEMOTIONAL SKILLS THROUGH VIRTUAL EXCHANGE (SOCIEMOVE: DESARROLLO DE HABILIDADES SOCIOEMOCIONALES EN INTERCAMBIOS VIRTUALES), Ref. PID2023-151087NB-I00, "Knowledge Generation Projects" call from the State Programme to Promote Scientific-Technical Research and its Transfer, within the framework of the State Plan for Scientific, Technical Research and Innovation 2021-2023, co-directed by Dr Ana Sevilla-Pavón and Dr Margarita Vinagre-Laranjeira from 01/09/2024 to 01/09/2027.
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Competing interest
Authors declare no competing interests.
Glossary
This framework outlines how to design learning activities and use CALL materials, which has four types of learning activities.
Students' preference in learning to target contexts that have deep and surface approaches.
The psychological framework that displays the serial interaction between stimulus organisms and response, where stimulus is the factor that facilitates action and influences an individual's state. The organisms mediate the effects of stimuli on responses in affective and cognitive states, and response is the reaction or behavior that the organism exhibits as a result of processing the stimulus.
A quantitative data analysis that has an asymmetrical nature where it predicts the target outcome based on the combinations of independent variables (case by case) that are responsible for generating outcomes. It follows the necessary logic and explores the must-have factors that must be present in the study to achieve the target outcome.
A quantitative advanced structural equation modeling approach that is variance-based and among the asymmetrical quantitative data analysis that focuses on the average impact of each variable on the other and follows fuzzy logic to explore the should have factors that should be present in the study to achieve the target outcome.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Antonietti, C; Schmitz, M-L; Consoli, T; Cattaneo, A; Gonon, P; Petko, D. Development and validation of the ICAP technology scale to measure how teachers integrate technology into learning activities. Computers & Education; 2023; 192, [DOI: https://dx.doi.org/10.1016/j.compedu.2022.104648] 104648.
Chen, W-C; Shih, Y-CD; Liu, G-Z. Task design and its induced learning effects in a cross-institutional blog-mediated telecollaboration. Computer Assisted Language Learning; 2013; 28,
Chi, MTH. Active-constructive-interactive: A conceptual framework for differentiating learning activities. Topics in Cognitive Science; 2009; 1, pp. 73-105. [DOI: https://dx.doi.org/10.1111/j.1756-8765.2008.01005.x]
Chi, MTH; Adams, J; Bogusch, EB; Bruchok, C; Kang, S; Lancaster, M; Levy, R; Li, N; McEldoon, KL; Stump, GS; Wylie, R; Xu, D; Yaghmourian, DL. Translating the ICAP theory of cognitive engagement into practice. Cognitive Science; 2018; 42,
Chi, MTH; Wylie, R. The ICAP framework: Linking cognitive engagement to active learning outcomes. Educational Psychologist; 2014; 49,
Coertjens, L. The relation between cognitive and metacognitive processing: Building bridges between the SRL, MDL, and SAL domains. British Journal of Educational Psychology; 2018; 88,
Colpaert, J. Editorial position paper: How virtual is your research?. Computer Assisted Language Learning; 2020; 33,
Demir, N; Kayaoğlu, MN. Multi-dimensional foreign language education: The case of an eTwinning project in Turkey. Computer Assisted Language Learning; 2021; 35,
Dooly, M; Sadler, R. If you don’t improve, what’s the point? Investigating the impact of a “flipped” online exchange in teacher education. ReCALL; 2019; 32,
Dul, J. Necessary condition analysis (NCA): Logic and methodology of “necessary but not sufficient” causality. Organizational Research Methods; 2016; 19,
Ellis, RA; Bliuc, A. An exploration into first-year university students’ approaches to inquiry and online learning technologies in blended environments. British Journal of Educational Technology; 2015; 47,
Ellis, RA; Bliuc, A-M. Exploring new elements of the student approaches to learning framework: The role of online learning technologies in student learning. Active Learning in Higher Education; 2017; 20,
Ellis, RA; Pardo, A; Han, F. Quality in blended learning environments–Significant differences in how students approach learning collaborations. Computers & Education; 2016; 102, pp. 90-102. [DOI: https://dx.doi.org/10.1016/j.compedu.2016.07.006]
Entwistle, N. Student learning, understanding, and implications for teaching; 2018; Elsevier: [DOI: https://dx.doi.org/10.1016/b978-0-12-805359-1.00012-7]
Forsblom, L; Peixoto, F; Mata, L. Perceived classroom support: Longitudinal effects on students’ achievement emotions. Learning and Individual Differences; 2021; 85, [DOI: https://dx.doi.org/10.1016/j.lindif.2020.101959] 101959.
Fredricks, JA; Filsecker, M; Lawson, MA. Student engagement, context, and adjustment: Addressing definitional, measurement, and methodological issues. Learning and Instruction; 2016; 43, pp. 1-4. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2016.02.002]
Freeborn, L; Andringa, S; Lunansky, G; Rispens, J. Network analysis for modeling complex systems in SLA research. Studies in Second Language Acquisition; 2022; 4, pp. 1-32. [DOI: https://dx.doi.org/10.1017/S0272263122000407]
Fütterer, T; Hoch, E; Lachner, A; Scheiter, K; Stürmer, K. High-quality digital distance teaching during COVID-19 school closures: Does familiarity with technology matter?. Computers & Education; 2023; 199, [DOI: https://dx.doi.org/10.1016/j.compedu.2023.104788] 104788.
Geisser, S. A predictive approach to the random effect model. Biometrika; 1974; 61,
Grau, MK; Turula, A. Experiential learning of telecollaborative competences in pre-service teacher education. Language Learning & Technology; 2019; 23,
Gruber, A; Canto, S; Jauregi-Ondarra, K. Exploring the use of social virtual reality for virtual exchange. ReCALL; 2023; 35,
Gunness, A; Matanda, MJ; Rajaguru, R. Effect of student responsiveness to instructional innovation on student engagement in semi-synchronous online learning environments: The mediating role of personal technological innovativeness and perceived usefulness. Computers & Education; 2023; 205, [DOI: https://dx.doi.org/10.1016/j.compedu.2023.104884] 104884.
Hafour, MF; De Felice, D; Kessler, G. A/synchronous telecollaborative digital media projects: Comparative effects on learners’ macro-and micro-level oral proficiency and intercultural competence. Computer Assisted Language Learning; 2023; [DOI: https://dx.doi.org/10.1080/09588221.2023.2269412]
Hair, JF, Jr; Howard, MC; Nitzl, C. Assessing measurement model quality in PLS-SEM using confirmatory composite analysis. Journal of Business Research; 2020; 109, pp. 101-110. [DOI: https://dx.doi.org/10.1016/j.jbusres.2019.11.069]
Hair, JF, Jr; Hult, GTM; Ringle, CM; Sarstedt, M. A primer on partial least squares structural equation modeling (PLS-SEM); 2021; SAGE Publications: [DOI: https://dx.doi.org/10.3926/oss.407]
Han, J; Geng, X. University students’ approaches to online learning technologies: The roles of perceived support, affect/emotion and self-efficacy in technology-enhanced learning. Computers & Education; 2023; 194, [DOI: https://dx.doi.org/10.1016/j.compedu.2022.104695] 104695.
Henseler, J; Ringle, CM; Sarstedt, M. A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the Academy of Marketing Science; 2015; 43,
Hiver, P; Al-Hoorie, AH; Mercer, S. Student engagement in the language classroom; 2020; Multilingual Matters: [DOI: https://dx.doi.org/10.21832/HIVER3606]
Kurek, M. Designing tasks for complex virtual learning environments. Bellaterra Journal of Teaching Learning Language Literature; 2015; 8,
Larsen-Freeman, D; Cameron, L. Complex systems and applied linguistics; 2008; Oxford University Press:
Luo, H; Gao, P. Intercultural learning through Chinese-American telecollaboration: Results of a song sharing project. Computer Assisted Language Learning; 2022; [DOI: https://dx.doi.org/10.1080/09588221.2022.2026405]
Luo, H; Yang, C. Twenty years of telecollaborative practice: Implications for teaching Chinese as a foreign language. Computer Assisted Language Learning; 2018; 31,
Luo, H; Yang, C. Pedagogical benefits of Chinese-American virtual exchange: A study of student perceptions. ReCALL; 2021; 34,
Marton, F; Säljö, R. On qualitative differences in learning: I—outcome and process*. British Journal of Educational Psychology; 1976; 46,
Mehrabian, A; Russell, JA. An approach to environmental psychology; 1974; The MIT Press:
Müller-Hartmann, A. (2005). Learning how to teach intercultural communicative competence via telecollaboration: A model for language teacher education. The American Association of University Supervisors, Coordinators and Directors of Foreign Languages Programs (AAUSC), pp. 63–84. http://hdl.handle.net/102015/69620
Niu, L; Wang, X; Wallace, MP; Pang, H; Xu, Y. Digital learning of English as a foreign language among university students: How are approaches to learning linked to digital competence and technostress?. Journal of Computer Assisted Learning; 2022; 38,
O’Dowd, R. Telecollaborative networks in university higher education: Overcoming barriers to integration. The Internet and Higher Education; 2013; 18, pp. 47-53. [DOI: https://dx.doi.org/10.1016/j.iheduc.2013.02.001]
O’Dowd, R. Virtual exchange: Moving forward into the next decade. Computer Assisted Language Learning; 2021; 34,
O’Dowd, R; Dooly, M. Exploring teachers’ professional development through participation in virtual exchange. ReCALL; 2021; 34,
O’Dowd, R; Sauro, S; Spector-Cohen, E. The role of pedagogical mentoring in virtual exchange. TESOL Quarterly; 2019; 54,
O’Dowd, R; Waire, P. Critical issues in telecollaborative task design. Computer Assisted Language Learning; 2009; 22,
O'Dowd, R. Jackson, J. Intercultural communicative competence through telecollaboration. The Routledge handbook of language and intercultural communication; 2012; Routledge: pp. 342-358.
Ou Yang, F-C; Lai, H-M; Wang, Y-W. Effect of augmented reality-based virtual educational robotics on programming students’ enjoyment of learning, computational thinking skills, and academic achievement. Computers Education; 2023; 195, [DOI: https://dx.doi.org/10.1016/j.compedu.2022.104721] 104721.
Payant, C; Zuniga, M. Learners’ flow experience during peer revision in a virtual writing course during the global pandemic. System; 2022; 105, [DOI: https://dx.doi.org/10.1016/j.system.2021.102715] 102715.
Qiu, X. Revisiting the cognition hypothesis: The impact of task complexity on L2 learner engagement in task performance in computer-mediated and face-to-face communication. Computer Assisted Language Learning; 2022; [DOI: https://dx.doi.org/10.1080/09588221.2022.2142245]
Rahimi, A. R. (2023). The role of EFL learners’ L2 self-identities, and authenticity gap on their intention to continue LMOOCs: Insights from an exploratory partial least approach. Computer Assisted Language Learning, 1–32 https://doi.org/10.1080/09588221.2023.2202215.
Rahimi, A. R., & Mosalli, Z. (2024). Exploring the direct and indirect effects of EFL learners’ online motivational self-system on their online language learning acceptance: The new roles of current L2 self and digital self-authenticity. Asian-Pacific Journal of Second and Foreign Language Education, 9(1). https://doi.org/10.1186/s40862-024-00266-0
Rahimi, A. R. (2024). A tri-phenomenon perspective to mitigate MOOCs’ high dropout rates: The role of technical, pedagogical, and contextual factors on language learners’ L2 motivational selves, and learning approaches to MOOC. Smart Learning Environments, 11(1) https://doi.org/10.1186/s40561-024-00297-7.
Rahimi, AR; Cheraghi, Z. Unifying EFL learners’ online self-regulation and online motivational self-system in MOOCs: A structural equation modeling approach. Journal of Computers in Education; 2022; 11,
Rahimi, AR; Sevilla-Pavón, A. The role of ChatGPT readiness in shaping language teachers’ language teaching innovation and meeting accountability: A bisymmetric approach. Computers and Education: Artificial Intelligence; 2024; 7, [DOI: https://dx.doi.org/10.1016/j.caeai.2024.100258] 100258.
Rahimi, AR; Sevilla-Pavón, A. A pathway from surface to deep online language learning approach: The crucial role of online self-regulation. Acta Psychologica; 2024; 251, [DOI: https://dx.doi.org/10.1016/j.actpsy.2024.104644] 104644.
Rahimi, AR; Sheyhkholeslami, M; Mahmoudi Pour, A. Uncovering personalized L2 motivation and self-regulation in ChatGPT-assisted language learning: A hybrid PLS-SEM-ANN approach. Computers in Human Behavior Reports; 2025; 17, [DOI: https://dx.doi.org/10.1016/j.chbr.2024.100539] 100539.
Reeve, J. (2012). A self-determination theory perspective on student engagement. In: Christenson, S., Reschly, A., Wylie, C. (eds) Handbook of research on student engagement. Springer https://doi.org/10.1007/978-1-4614-2018-7_7.
Rienties, B; Lewis, T; O’Dowd, R; Rets, I; Rogaten, J. The impact of virtual exchange on TPACK and foreign language competence: Reviewing a large-scale implementation across 23 virtual exchanges. Computer Assisted Language Learning; 2020; 35,
Sevilla-Pavón, A; Nicolaou, A. Artefact co-construction in virtual exchange: ‘Youth Entrepreneurship for Society’. Computer Assisted Language Learning; 2020; 35,
Schmidt, SJ. Teaching AND reaching our students from a distance. Journal of Food Science Education; 2020; 19,
Skinner, EA; Kindermann, TA; Furrer, CJ. A motivational perspective on engagement and disaffection. Educational and Psychological Measurement; 2008; 69,
Skinner, EA; Pitzer, JR. Christenson, S; Reschly, A; Wylie, C. Developmental Dynamics of Student Engagement, Coping, and Everyday Resilience. Handbook of Research on Student Engagement; 2012; Boston, MA, Springer:
Smarandache, IG; Maricutoiu, LP; Ilie, MD; Iancu, DE; Mladenovici, V. Students’ approach to learning: Evidence regarding the importance of the interest-to-effort ratio. Higher Education Research & Development; 2021; 41,
Stone, M. Cross-Validatory choice and assessment of statistical predictions (with discussion). Journal of the Royal Statistical Society Series b: Statistical Methodology; 1976; 38,
Sulis, G; Philp, J. Exploring connections between classroom environment and engagement in the foreign language classroom. Student Engagement in the Language Classroom; 2020; Channel View Publications: pp. 101-119. [DOI: https://dx.doi.org/10.2307/jj.22730722.11]
Sun, JC; Rueda, R. Situational interest, computer self-efficacy and self-regulation: Their impact on student engagement in distance education. British Journal of Educational Technology; 2011; 43,
Sun, PP; Zhang, LJ. Investigating the effects of chinese university students’ online engagement on their EFL learning outcomes. The Asia-Pacific Education Researcher; 2024; 33,
Ware, P; Kessler, G. Telecollaboration in the secondary language classroom: Case study of adolescent interaction and pedagogical integration. Computer Assisted Language Learning; 2014; 29,
Wu, S. (2021). Unpacking themes of integrating telecollaboration in language teacher education: A systematic review of 36 studies from 2009 to 2019. Computer Assisted Language Learning, 1–23. https://doi.org/10.1080/09588221.2021.1976800.
Yang, R. The use of questions in a synchronous intercultural online exchange project. ReCALL; 2017; 30,
Zakariya, YF; Nilsen, HK; Goodchild, S; Bjørkestøl, K. Self-efficacy and approaches to learning mathematics among engineering students: Empirical evidence for potential causal relations. International Journal of Mathematical Education in Science and Technology; 2020; 53,
Zhang, Z. V., & Hyland, K. (2023). The role of digital literacy in student engagement with automated writing evaluation (AWE) feedback on second language writing. Computer Assisted Language Learning, 1–26 https://doi.org/10.1080/09588221.2023.2256815.
© The Author(s) 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.