The intercultural communication competence improvement for pre-service CSL teachers: A blended learning method based on SVVR

2.1Training methods of pre-service foreign language teachers’ intercultural communication competence

Scholars agree that intercultural communicative competence should be integrated into the curriculum content of pre-service training for foreign language teachers [23–25], and most of them use traditional teaching methods to improve the professional skills and intercultural communication competence of foreign language teachers. However, as Hoa and Thi [26] point out, because intercultural competence has its own dimensions and developmental framework, it necessitates a distinct intercultural language learning and teaching (IcLLT) approach. It means that learners should actively participate in social interaction and critical reflection during the five steps of “construction,” “connection,” “reflection,” “interaction,” and “extension.” According to Benattabou (2020), the effective implementation of “critical incident” exercises, as well as the use of role-play and audiovisual media tools, can raise learners’ awareness of the potential risks associated with cross-cultural encounters [27].

In the study of intercultural communication competency training approaches for pre-service CSL teachers, Peng (2013) proposes an interactive pedagogy in the classroom between international students and pre-service CSL teachers [28]. Zheng and Li [29] propose a reflective intercultural teaching model that encourages students to actively explore various cultural phenomena and use archival portfolios to record the process of cultural research conducted by students, which can significantly enhance intercultural communication competence at the affective and behavioral levels [29]. Unfortunately, the majority of the teaching styles presented in these few studies are focused on offline classrooms, and more research on how to assist CSL teachers in improving the effectiveness of intercultural communication competence is still needed.

2.2ICT application in pre-service foreign language teacher training

Recently, governments all around the world have stepped up preparations and actions in response to the New Coronary Pneumonia epidemic [30], and social media use has significantly impacted people’s mental health [31]. It is increasingly noteworthy that this apparent trend towards IT integration is also present in the training of pre-service ESL teachers. Technologies such as web-based virtual communication and simulation experiences [32], ICT-supported audio training sessions [33], and the use of MICT (Mobile Information and Communication Technology) technology [34] have proven effective for preservice teacher training, and MALL (Mobile-Assisted Language Learning) training has even positively impacted preservice teachers’ cognition and used [35]. The traditional C-BL (Computer-Blended Learning) model, on the other hand, primarily presents instructional content through online videos and offline courses. Most students find it challenging to experience the real-world scenarios faced during professional training in such a one-sided information transmission and unproven learning environment, which impairs their judgment and actual processing performance [36].

There has not been much research on pre-service CSL teacher preparation that incorporates technology. Although the CoCAR (Cooperation-based cognition, action, and reflection) approach, which emphasizes cognition, action, and reflection, is seen to be suitable for developing pre-service CSL teachers’ online synchronous teaching skills, and can be used to enhance instructors’ ICT proficiency in general training programs [37]. However, Lin et al. (2014) discovered that a significant hurdle for pre-service CSL teachers is a lack of time to create technology-driven pedagogies and activities for ICT uptake [38]. On the other hand, recent related research has demonstrated that using intelligent teaching tools like Moso Teach (MT) and Rain Classroom (RC) (a new smart teaching tool that combines WeChat and other social media platforms) to develop instructional design techniques can improve the intercultural communication competency of preservice CSL instructors [39].

Although prior research on the development of foreign language teachers’ intercultural communicative skills, with English teachers as the core group, and its integration with information technology, has produced considerable outcomes; However, the research on the pre-service CSL teachers still receives little attention.

2.3SVVR application in pre-service foreign language teacher training

Among the different types of ICT technology applications, Spherical Video-based Virtual Reality (SVVR) has the advantage of providing highly immersive and highly interactive learning environments that can compensate for the shortcomings of other information technologies. It has been shown that its application in teaching practice offers new possibilities for the development of students’ problem-solving skills, provides richer contexts for students, and makes the learning process more interesting and interactive [40]; and it is an information technology that can facilitate effective learning for a wide range of learner types and ages [41–44]. According to some studies, SVVR, because of its ability to help learners experience being more immersed in authentic socio-cultural environments and English language practice environments, has had a more positive effect on learners’ spoken English, motivation, and critical thinking skills, and has helped to reduce their English language learning anxiety, and this technology has proven to be effective in improving the oral performance of EFL teachers [45] and the perception of learning in terms of being effective [46]. Moreover, SVVR training using dual-loop technology can change students’ social and behavioral engagement, which may help learners achieve good writing learning outcomes [47]. The experiences brought about through SVVR technology can help improve EFL writing learners’ contextual knowledge and understanding and help learners make constructive intrinsic connections between their personal experiences and their writing topics, thus helping to improve the quality of their writing [48]. However, among these existing studies, there is still a lack of research on SVVR being applied to pre-service CSL teachers.

3.1Course material VRization and algorithm

Virtual reality systems necessitate simulations using personal computers and workstations, with the computer screen serving as a window through which the user can observe the virtual environment [58]. To make it easier for teachers to develop SVVR course material independently, we standardized and encapsulated the course material development process (shown in Fig. 1) and identified the underlying algorithms. The spherical 3D mapping to 2D planes process consists of two basic steps: 3D space mapping and 2D plane display [59].

Fig. 1

Schematic diagram of spherical 3D mapping to 2D plane.

Drawing Earth in a cosmic star environment is the first step in 3D space drawing; the drawing method includes the following steps: (1) draw a 3D sphere: In the application, create a cosmic star background and draw a 3D sphere on top of it; (2) sphere mapping: map on the 3D sphere; (3) add sphere rotation: set the 3D space light source so that the 3D sphere rotates cyclically, similar to Earth rotation.

By using particular 3D space drawing techniques and 2D plane display techniques, the point-to-plane specific coordinates of the sphere are associated with the 2D plane display, thereby realizing the coordinate mapping of the 3D sphere to the 2D plane, achieving better 3D visual effects, and providing learners with a more novel interactive experience. These standardized templates help to control the cost of developing a large number of subsequent course materials and are simple to implement.

The algorithm and equation for the conversion of sphere coordinates to two-bit planar pixel coordinates [60] are as follows:

By simplifying, it becomes:

The M matrix is a 4*4 matrix, and in the same plane with zw = 0, the M matrix can change to a 3*3 matrix, expressed as follows.

Combining the two formulas above, H is the single response matrix, and the two sides of the H matrix correspond to the two related images. It demonstrates that using the H single response matrix, the points of one plane in 3D space can be mapped (encoded) into the coordinates of the two cameras’ related images, thereby realizing the content production and projection transformation work of SVVR.

The above formulas (1) to (4) map matching point pairs on two images (binoculars), and they can also be used with other matrices to perform functions like scaling and rotation.

3.2Course authoring principles and design

The curriculum for this study was heavily influenced by the Cone of Experience and Cone of Learning theory developed by Edgar Dale, the father of modern media in education (see Fig. 2). According to this theory, the sources of experience can be either indirect or direct, and there are significant differences in the learning outcomes obtained by different learning styles, with those that rely solely on self-reading and teacher instruction being the least effective, and those that rely on hands-on participation or teaching to others being the most effective. It implies that active learning is more effective than passive learning and that active participation and practice are effective learning methods [61]. Dale’s theory can assist teachers in identifying teaching resources and activities.

Fig. 2

Cone of Experience.

According to Dale’s proposition, we refer to the principles of TVET (technical and vocational education and training) curriculum development [62], and we combined these principles with the characteristics of the SVVR technology application to create the following targeted and optimized design for training curriculum development:

First, as the foundation for the training curriculum design, we conducted an in-depth study of international pre-service intercultural communication courses for Chinese language teachers. Second, following the constructivist theory of individual cognitive differences [63], we collected the feedback and evaluation of learners and teachers on such courses in universities with volunteers via one-on-one interviews and surveys and then extracted the features of the database composed of this feedback and evaluation via feature analysis and determined the input of follow-up samples. Third, we used the K-mean judgment method to identify the characteristics of the factors that influenced previous course training; Fourth, we collected primary data from different learners and provided them with targeted learning resources using web 2.0 personalized displays. Targeted learning resources to ensure that learners’ knowledge acquired during the learning process is consistent with their prior experience, allowing them to construct the target knowledge and ability more quickly; Fifth, we emphasize the visual information perception function by using a standardized and unified picture style to create an immersive picture experience for learners, and TEVT-related formats of instructional materials can be called, together with video sequences based on SVVR technology with binocular panoramic input, to complete the preservation of high-quality audio and video digital files, which are converted to the sphere via built-in projection. It also uses sound effects to help learners gain contextualized experiences, improve knowledge and digestion, and achieve teaching objectives.

Furthermore, during the experimental phase of this training course, the following design concepts are highlighted.

We use SVVR to interact with learners’ existing knowledge, experience, and accurate information through the virtual reality display effect to realize the continuous cycle of learners from situation-action-reflection, and finally guide learners to generate new practical knowledge in the process of solving practical problems, using the above design idea. Finally, instructors direct students to independently create new practical knowledge while resolving real-world issues.

3.3Training processing and modules

Training is a crucial part of pre-vocational education, and its main objectives are to improve learners’ employability and practical problem-solving skills. By computer simulation of various human-computer interaction scenarios and editing works with multiple perceptual characteristics, SVVR can help learners immerse themselves in a 3D simulation environment [64], and it can also significantly reduce learners’ learning time through multiple spatial substitutions and improve the completion of intercultural interactive training. Through multiple spatial substitutions, it can also significantly reduce learner learning time and increase the completion of cross-cultural communication interaction training.

We have gathered some of the audio and video materials for the cross-cultural communication training course for pre-service CSL teachers through traditional photography techniques, followed by the processes of spherical splicing, rotation, drawing, flat area compression, and encapsulation. These materials include daily communication scenes in Chinese cultural context and interactive scenes in different cultural contexts that may be taught by future international Chinese teachers, among others. Through a large number of software iterations and reduced hardware investment, the virtual experimental classroom effectively solves the dependence of intercultural communication practical training on physical conditions such as venue and equipment. It can continuously enrich course teaching resources according to changing reality, and learning needs to meet the pre-service training needs of international Chinese teachers.

The following typical practical training models (see Table 1) can be standardized through the central core process of SVVR processing, with different models depending on different audio and video materials and practical training purposes. In order to boost scalability and finish the entire process of configuring digital course materials and presentation restoration to match the actual teaching needs, the orange area module in Fig. 3 may code and decode these various sorts of models.

Fig. 3

Main core processes of SVVR processing

4.1Sample collection and multimodal recognition

Mehrabian proposed the PAD three-dimensional theory of emotion as a model of dimensional observations. According to the model, emotions have three dimensions: pleasure, arousal, and dominance [65]. P stands for Pleasure-displeasure, which denotes the subject’s positive and negative emotional state; A stands for Activation-nonarousal, which indicates the subject’s level of neurophysiological activation; D stands for Dominance-submissiveness, which indicates the subject’s state of control over the situation and others, is the state in which the individual’s emotions are controlled and dominated; spontaneous emotions from within are dominant emotions, such as anger, and passive emotions from without are submissive emotions, such as fear. We believe that the values of these three dimensions, like other factors in the past, influence changes in learning efficiency, referred to as PADL (see Fig. 4).

Fig. 4

PADL relationship diagram.

We conducted relevant experimental design, data collection, and feature training based on PADL to facilitate the subsequent discovery of patterns between affective changes and learning. Gender, age, English level, experience abroad, and relevant course learning experience were among the personal information of learners collected for the experiment; during the experiment, the data on the learners might be automatically updated to reflect their level of autonomous study.In order to better set up the a priori data [66],we categorised the different forms of data after identifying the aforementioned indicators (see Table 2).

ICED: Learner evaluations add significant value to the study’s degree of effectiveness, strengths, and weaknesses, as well as subsequent improvements and refinements.

TSD: RNN (LSTM) and CNN are methods that exploit the semantics between words for text represented as a sequence of word embeddings. ASTM takes as input each cell with a word embedding and the previous cell’s hidden state, calculates the output, and updates the hidden state of the next cell. The software keeps track of the semantics of the previous words using the fundamental operations. CNN computes the local contextual characteristics between consecutive words using convolutional techniques. The collected features are subsequently merged using the average or top pooling layer for future sentiment categorization. We recommend using BERT and GPT-3 to assist with the tasks mentioned above.

ASD: We currently use the HMM algorithm for audio sentiment recognition, a usual method capable of capturing dynamic features of sequence data in real-time. Audio emotion recognition has also made extensive use of support vector machines.

ISD: Using the image features, the manually extracted image features are first combined, and the data is then fed into a support vector machine to train the classifier. Finally, the classifier and feature extractor are optimized based on deep learning with the corresponding loss functions such as cross-entropy loss for end-to-end connectivity.

VSD: A two-stage pipeline is utilized to extract visual/audio information and train the classifier in order to determine video sentiment. SVM, GMM, HMM, Dynamic Bayesian Network (DBM), and Conditional Random Time are all included (CRF).

Based on the five approaches described above, we perform multimodal hybrid data training and recognition, including subjective data obtained from the analysis of teaching experts and teaching resources and objective data collected dynamically in real-time. A neural network approach to emotion recognition achieves an end-to-end approach to video emotion recognition [67]. The objective requirements for teaching are effectively predicted and verified in real-time, while SVVR monitors the performance of the teacher and learners during the teaching process (see Fig. 5).

Fig. 5

Multimodal hybrid training and recognition.

4.2Learners’ emotion capture and correspondence training

Since six factors, behavioral flexibility, interactive relaxation, interactive respect, information skills, identity maintenance, and interaction management, constitute cross-cultural validity [68], learners’ interactive performance in cross-cultural communication is critical. The emotional characteristics of learners and emotion capture can be utilized to provide feedback on the conduct and effectiveness of virtual intercultural communication engagements. We created multiple feature data subsets by extracting and analyzing participants’ emotional features, referencing existing datasets, and using each feature subset to train and test learners while also updating the emotional feature subsets, yielding a learner-learning emotion recognition model (see Fig. 6).

Fig. 6

Schematic diagram of the research method of emotion recognition features.

Previous research has focused on the affective dimension of intercultural communicative competence, highlighting communicators’ emotions or feelings when interacting with people from different cultural backgrounds or in specific situations [69]. As a result, the changes and frequency of emotions captured in virtual scenarios can help us determine learners’ intercultural communicative competence. We discovered through video data that when learners had positive interactional behaviors, they usually showed positive emotions in online learning, such as pleasure, friendliness, enthusiasm, and sympathy, and some were even able to sustain these emotions for a more extended period. In contrast, when learners’ interactional interactions were negative and appeared passive, they tended to show anger, irritability, and tenacity.

Additionally, we found that customized instruction could be carried out online, and preliminary attempts were made in addition to recording learners’ emotions. Currently, a large number of emotion training resources are freely available on the Internet. The following table’s numerous emotion model libraries (see Table 3) are used to train artificial intelligence before further classifying and sampling the emotional traits of students in the TEVT course [70], which can be easily used for project tracking and business judgment in the subsequent learning process and to provide constructive feedback on learner learning and development [71].

4.3Effectiveness of SVVR blended learning method

We randomly assigned the experimental (30) and control (30) groups of 60 volunteers. We used the online method that learners have adapted to carry out most steps of the teaching experiment since 2020 because the volunteers were distributed in different cities, taking into account the cost of transportation and other factors, and referring to the delivery format of other similar training courses (except for SVVR). The same professional instructor with over ten years of teaching experience taught both groups. The experimental group used the SVVR blended learning method, while the control group used traditional teaching methods, such as instructor-led lectures. Concerning the lesson time allocation structure of previous similar training courses, and keeping in mind that excessive use of virtual reality scenes may impose visual cognitive load on students [72], we designed the experimental course as follows: Both groups used the instructor-led teaching method in the essential knowledge section (8 lessons, 30 minutes each); both groups used the instructor-led teaching method in the application ability section (4 lessons, 25 minutes each). The control group used the instructor’s lecture method (4 hours, 25 minutes each), while the experimental group used the SVVR learning method.

Both groups of learners took the same learning test at the end of the course, which consisted of a paper test on the fundamentals and a virtual simulation game-based test on their application skills. The testing process and timing are identical, and the final score is derived using the combined results of the two tests. The teacher created the paper test based on the syllabus and teaching content, and the questions were divided into four types: multiple choice, judgment, terminology, and exposition. The test lasted 30 minutes and was worth 50 points. Multiple scorers scored the expository questions, and three professional teachers were invited to do so. The Intraclass Correlation Coefficient (ICC = 0.856 > 0.8) was calculated. We chose the interactive part of the intercultural scenario for the test because it is an experimental module for intercultural communication training for foreign language teachers developed by a university teaching team (it has been used in actual teaching and approved by the competent educational department). The above two items scored a total of 100 points. Referring to the current method of data analysis in recent studies [73–75], the results of the analysis of the test data for both groups of participants are shown in Fig. 7.

Fig. 7

Comparison of test data between the traditional learning method and SVVR hybrid learning method.

The Independent sample t-test results show that different methods have significant differences in learner performance scores (t = –4.173, df = 29, p < 0.01); the traditional method’s score is significantly lower than that of SVVR (MD = –5.583). There was no significant difference between the experimental and control groups in the fundamental knowledge test (t = –1.037, df = 29, p > 0.05); however, in the application ability test, the experimental learners scored significantly higher than the control learners (t = –6.185, df = 29, p0.01, MD = 11.166), indicating that they had more vital intercultural communication skills. The experimental results show that using SVVR technology can help pre-service CSL teachers improve their intercultural communication skills to some extent. Simultaneously, the system can provide an advanced tool for testing pre-service CSL teachers’ intercultural communicative competence and a novel idea for quantifying the SVVR online test.

To further understand the experience and feelings of the participants of the SVVR blended learning method, we launched recruitment for voluntary semi-structured interviews and received responses from 10 learners. Eight of these learners indicated positive attitudes toward this learning method, “I think this learning course makes me experience freshness and authenticity instead of the boring feeling that comes from learning knowledge all the time, and I think this is very prominent.” (Male, 23 years old) “I found this learning method very efficient because it helped me to feel a different cultural scenario and put myself in it in a very short period and engage in the interactive process. I think it will be beneficial for my future teaching work that I may do overseas.” (Female, 24 years old) Moreover, the teacher also expressed the view but also suggested the direction of improvement, “I think this should be the future direction, you know, the use of AI in education and training has become more and more popular, especially this kind of application that can cross time and space, it can provide much help to teachers and students. However I think it is not intelligent enough at the moment; maybe many your design ideas have not been reflected yet, or maybe you should collect more material on different scenarios of overseas learners learning Chinese.” (Female, 42 years old)

According to contextual learning, when learners learn in a realistic and interactive context, they can rationally and meaningfully interpret the acquired knowledge and skills [76]. The blended learning method based on SVVR provides learners with an immersive 3D simulation environment for cross-cultural interactions, which can stimulate their interest in participation and bring situational and interactive experiences, thus facilitating their understanding and application of knowledge, which meets the criteria of meaningful learning and enables learners in the experimental group to more easily self-construct practical knowledge related to cross-cultural communication.