Teacher-Student Interaction for English-Medium Instruction (EMI) Content and Language Learning and the Effects of Implementing Multimodal Input of Classroom Interaction: University Students’ Perceptions

The discourse of interaction, Initiation-Response-Feedback (I-R-F) gained its popularity by fostering discussion between the teacher and learners but received critics for developing interactive communication in a controlled manner. With an attempt to provide pragmatic implications for EMI instructors, the study probed into perceptions of students (n=42) from different majors on how two pedagogical approaches (i.e., the I-R-F and the multimodal input of classroom interaction) differed in one EMI Cross-cultural Communication course in one middle-ranked university in Taiwan. The comparative, quasi-experiment research firstly investigated these students’ perceptions of teacher-student interaction for EMI content and language learning and secondly compared their perceptions on the implementation of the multimodal input of classroom interaction against the conventional baseline, the I-R-F. Both quantitative (i.e., survey) and qualitative (i.e., post-lesson student reflection journals and audio recordings) research methods were used. The survey results showed that these students were inclined to engage in extensive and substantial verbal output, expressed the importance of teacher-student interaction for learning the content of the course, and expected chances of lengthy verbal output and corrective feedback from the instructor. The results from students’ journals yielded that the instructor’s use of the multimodal input of classroom interaction significantly outperformed the use of the I-R-F in the categories of effectiveness, the level of student engagement, and the effectiveness of helping them learn the content. Ultimately, it was found that the use of the multimodal input of classroom interaction had triggered their higher-order of cognitive processing more (i.e., analyzing, evaluating, and creating).


Introduction
English-Medium Instruction, henceforth EMI, refers to that non-language academic subjects are taught through the medium of English in regions where English does not serve as the primary language of communication (Jenkins, 2017;Llieva & Peak, 2017). Although English is recognized as a foreign language in Taiwan (Chen & Tsai, 2012), with a vision of Taiwan's 2030 goal to become a bilingual nation commissioned by National Development Council, Taiwan is aimed to raise the overall national competitiveness and English proficiency among the public (ROC National Development Council, 2020). One of the various working policies from the Ministry of Education (MOE) is putting EMI at all educational levels into a more full-fledged practice to improve the academic environment quality and increase student mobility for education, both inward and outward. In Higher Education (HE) settings, EMI courses across the country have evolved at a faster rate so that students from different linguistic and cultural backgrounds can study in Taiwan's tertiary institutions. As of 2018, 65,000 international students enrolled in Taiwan's HE institutions, and more than 171 EMI programs were offered (Puspitasari et al., 2020). Notably, as of 2021, approximately 500,000 students enrolled in 17,000 EMI courses (Xie, 2021). A rapid escalation in the number of programs and courses delivered in English has been seen.
in Bloom's taxonomy (Anderson & Krathwohl, 2001). Despite the attempt of previous efforts to identify dialogic pedagogy (Jones & Chen, 2016) and dialogic teaching (Reznitskaya, 2012), there is little research on the methods to promote classroom interaction (An & Thomas, 2021;Qiu & Fang, 2022). Furthermore, viewing the ongoing struggles and concerns from both instructors and students in EMI contexts worldwide, evidence-based research on EMI pedagogical practices is severely under-researched (Chuang, 2015), not to mention the ensuing theoretical and practical implications. Thus, the idea grounded in the study is that EMI instructors and students should optimally work in a mutually supportive manner to yield instructors' pragmatic and effective strategies.

Interaction Hypothesis & the Classroom Discourse of Initiation-Response Feedback
Although previous studies have identified various dilemmas and challenges in the practice of EMI worldwide, interaction-based pedagogies have been proven to be effective as they facilitate students' comprehension and lecturing communication (Flowerdew et. al., 2000;Kym & Kym, 2014;Suvinitty, 2010). Long's (1996) Interaction Hypothesis emphasizes the significance of negotiation of meaning; namely interaction can facilitate language learners' modifications of their own input and/or output. Interacting with others while learning has been highly supported in scholarship (Gass & Mackey, 2007;Okita, 2012), as interaction can assist learners to organize thoughts, reflect on understanding, and find gaps in reasoning effectively. Byun et al. (2010) demonstrated the lack of teacher-student interaction was detrimental to EMI learning. Yeh (2014) stated that teacher-dominated EMI lecturing style caused students' dissatisfaction, and Tsou and Chen's (2019) findings revealed that the Native English-Speaking Teachers (NESTs) were favored EMI instructors because most NESTs were encouraging and interactive, and adopted a less textbook-bound activity approach.
In terms of interaction-based lecturing and communicating approaches, Sinclair and Coulthard's (1975) proposed the classroom discourse of Initiation-Response Feedback (I-R-F), describing a triadic structure of which classroom discussion occurs when the teacher initiates an interaction, usually by asking a question, the learner responds, and then the teacher gives feedback as illustrated in Figure 1. The I-R-F is a useful framework of the exchange of information and classroom interaction but has been criticized for creating a controlled and teacher-dominated form of meaning negotiation (Alexander, 2010;Fisher, 2007;Lyle 2008).  (Sinclair & Coulthard, 1975)

Enhancing Language Learning with Multimodal Input
The didactic use of multimodal input refers to the use of multiple forms of input including visual, auditory, sensory, and textual modes such as visuals, images, YouTube videos, diagrams, demonstrations, actions, inquiry/discovery activities and practice, as shown in the red oval frame of the Multimodality-Entextualization Cycle (Lin, 2015) in Figure 2. The multimodal input of instruction was reported to enhance foreign or second language (L2) learning (Liu, Jang, & Roy-Campbell, 2018). Though the use of computer tools facilitated class interaction, the significance of the instructor's multimodal interaction with multiple forms of input was the key to shape class discourse and foster students' language development (Kim, 2021). Maiullo (2022) suggests that the activities employing multimodal input of visual, auditory, and other sensory materials can increase teacher-student interaction and better equip students with comprehension and communication skills. The multimodal nature of the audio-visual input from subtitles and captions in films were effective in learning foreign languages (Birulés-Muntané, & Soto-Faraco, 2016). Montero Perez (2022) also concluded that the audio-visual input and on-screen text were conducive to the development of vocabulary, grammar, and listening in L2. Though previous studies attempted to generalize the positive effects of multimodal input on language learners, little was done in EMI contexts.

Study Purpose and Research Questions
Owing to the scarcity of empirical evidence addressing students' views on interaction-based didactic practice and the instructor's use of multimodal input in EMI classrooms, the single-group comparative, quasi-experiment study incorporated multimodal interaction into EMI content teaching to firstly evaluated EMI students' beliefs on teacher-student classroom interaction for content and language learning and secondly compared these students' perceptions on the implementation between the conventional baseline, the I-R-F classroom discourse of interaction and the multimodal input of classroom interaction. The research sought answers to the following questions: 1. What are the students' perceptions on teacher-student interaction for learning EMI Cross-cultural Communication content and English language? 2. How do the students' perceptions differ between the I-R-F classroom discourse and the multimodal input of classroom interaction?

Context and Participants
The target participants were 42 local Taiwanese undergraduate students enrolled in an elective, non-language academic EMI subject, entitled Cross-cultural Communication, in one middle-ranked university in Taiwan. The course met 3 hours weekly for consecutive 18 weeks, and the student body comprised of students aged 19-22 years old (M=20.7) from the Departments of English (n=27, 64%), International Business (n=10, 24%), Regional and Social Development (n=2, 0.5%), Early Childhood Education (n=1, 0.25%), Digital Content and Technology (n=1, 0.25%), and Creative Design and Management (n=1, 0.25%), with male (n =12, 28.6%) and female (n =30, 71.4%). Students' English proficiency levels were TOEIC 880-950 (n=5, 15%), TOEIC 750-879 (n=18, 55%), TOEIC 550-749 (n=8, 19%), TOEIC 350-549 (n=1, 3%), and below TOEIC 349 (n=1, 3%). 37 (88%) out of the 42 students had previous EMI experience prior to attending this course: over 20 EMI courses (n=14, 33%), 10-19 EMI courses (n=7, 17%), 1-9 EMI courses (n=16, 38%), and only 5 of them (1.2%) had no prior EMI experience. Table 1 displays the demographics of the study participants. The EMI course was taught by a local Taiwanese who had a high English proficiency as he obtained his Master's in Teaching English to Speakers of Other Languages (TESOL) and a Ph.D. in Cross-cultural Education, both from the United States. The purpose of the course was to introduce learners to crucial issues, concepts, and working knowledge of cross-cultural communication with an emphasis on an active learning process and a student-centered interactive discussion style. Since the study aimed to compare students' perceptions on the implementation between the conventional baseline, the I-R-F classroom discourse of interaction and the multimodal input of classroom interaction, the instructor adopted the I-R-F classroom discourse of interaction for the first three weeks (i.e., 9 hours) of online instruction. The course was conducted online mainly due to the Covid-19 pandemic during the beginning period of the research. To reduce the effects that different settings (i.e., online and face-to-face) could have on the study results, the course was taught by the same instructor, and the two different pedagogical approaches (i.e., the I-R-F and the multimodal input of classroom interaction) were clearly communicated with the instructor and rigidly followed in lesson planning. The example of instructor's initiation (I), students' responses (R), and the instructor's feedback (F) is displayed in Appendix A.
After the use of the I-R-F classroom discourse of interaction in online setting, the same instructor and 42 students met in person. In the following three weeks (i.e., 9 hours), the instructor incorporated the multimodal input (e.g., videos, images, hands-on activities, simulations, or interactive learning activities) by designing a new construct of classroom interaction, Induce-Present-Reinforce-Debrief (I-P-R-D), a continuing cycle with four arrows representing four-phase pedagogical sequences into EMI content instruction as in Figure 3. In Phase 1: Warm-up (Induce), the instructor incorporated multimodal input such as videos, images, critical incidents, analogy, hands-on activities, or interactive learning activities to initiate students' participation and arouse their interest in teacher-whole or student-group class discussion before introducing a new concept. In Phase 2: Introduce a new concept (Present), the instructor connected students' inputs in the introduction of the new concept to build a conceptual common ground via teacher talk (monologues). In Phase 3: Reinforce the new concept (Reinforce), the instructor further deepened students' understanding of the content by employing multimodal activities such as critical incidents, matching, worksheets, simulations, practice, or interactive learning activities again in which the students engaged rigorously and acquired efficient learning experiences. In Phase 4: Debrief the learning target (Debrief), the instructor concluded what had been learned and achieved via teacher talk (monologues). Appendix B displays the sample course design using the I-P-R-D sequence of classroom interaction along with students' responses.

Questionnaire
A questionnaire containing a set of questions with multiple rating scales enables researchers to collect information from a number of respondents (Gillham, 2008). Thus, to answer RQ1 about students' perceptions on classroom interaction for content and language learning in an EMI Cross-cultural Communication course, a questionnaire comprising of three sections was employed. Part A contained six demographic information items inquiring students' names, ages, department of major, English proficiency test results obtained, previous EMI experience prior to attending this course, and the reasons for enrolling in this course. The following two sections adopted the existing questionnaires from An and Thomas (2021), designed to investigate how students perceived the role of interaction for subject knowledge learning and language learning in EMI contexts. Part B contained six items, generated by researching related literature of teacher-student interaction with the mean inter-item correlation (.21), to enquire about respondents' beliefs on teacher-student interaction for learning the content in the Cross-cultural Communication course. Part C included four items, adapted from existing items in validated questionnaires, to explore their beliefs on the role of interaction for English language learning in the course. A 5-point Likert scale that ranged from strongly disagree to strongly agree was used for each item.

Post-lesson Student Reflection Journals
Since reflection journals with open-ended questions further yield respondents' narrative insights (O'Connell & Dyment, 2011), two post-lesson reflection journals with six open-ended questions were issued, one immediately after the 3-week online instruction and the 3-week face-to-face instruction respectively, to elicit students' actual learning experience (RQ2). Both journal reflection questions were validated by two experienced researchers.

Student Reflection Journal for Online Instruction
Before issuing the student reflection journal questions, the I-R-F classroom discourse was clearly explained. The questions in the students' reflection journal for the online instruction included their reflections on 1) their perceptions on the instructor's use of the I-R-F sequence; 2) whether this sequence led to more or less teacher-student interaction; 3) the specific activity inspiring them to participate more; 4) whether the sequence was effective in helping them learn course content through English; 5) their challenges to participate in classes; 6) any suggestions to the I-R-F sequence.

Student Reflection Journal for Face-to-face Instruction
Before administering the student reflection journal questions, the I-P-R-D framework was also clearly elaborated. The questions for the students' reflection journal on the I-P-R-D framework included 1) their overall evaluation on the instructor's use of the I-P-R-D sequence as compared to the use of the I-R-F sequence; 2) whether the sequence led to more or less teacher-student interaction as compared to the instructor's use of the I-R-F sequence; 3) the specific activity inspiring them to participate more; 4) whether the sequence was a more effective way to help them learn course content through English against the use of I-R-F sequence; 5) their challenges to participate in classes as compared to those in classes of the I-R-F sequence; 6) any suggestions to the I-P-R-D framework.

Audio-Recordings
To increase the credibility and validity of the analyses on the collected data like complex human behaviors or perceptions, triangulation is an appropriate research method to cross-examine and provide in-depth standpoints (Altrichter et al., 2008;Cohen, Mansion, & Morrison, 2000). Thus, to further triangulate with the results from the two post-lesson student reflection journals, the first three weeks of online instruction and the following three weeks of face-to-face instruction were audio-recorded to analyze the type and number of EMI instructor's questions and students' responses.

Data Collection Procedures
After obtaining the approval from students' voluntary participation in the study, the first three consecutive online and face-to-face classes were audio-recorded, and students were invited to fill out the questionnaire of teacher-student interaction and to complete their first reflection journal on the 3 rd week after the online instruction ended. The second reflection journals were collected on the 6 th week after the face-to-face mode of instruction had been completed. The data collection procedure is demonstrated in Figure 4. In total, 42 student questionnaires, 42 post-lesson reflection journals of online instruction, and 42 post-lesson reflection journals of face-to-face instruction were collected. In total, both the questionnaires and two reflection journals were returned with a response rate of 100%.

Data Analyses
The data analyses included the followings: (1) The descriptive statistics: The mean, standard deviation, and the percentage of students' responses in the questionnaire were reported in three categories: a) strongly agree and agree, b) neutral, and c) strongly disagree and disagree).
(2) A hybrid process of inductive and deductive thematic coding analysis for the two post-lesson student reflection journals: This hybrid coding analysis recognizes the patterns of meaning within the data of social phenomenology such as experiences, views, and perceptions from reflection journal responses by using inductive coding and employing existing theoretical framework to identify preconceived themes in the data with deductive coding (Fereday & Muir-Cochrane, 2006) with the following six steps (Braun & Clarke, 2006).
Step 1 is to familiarize the data, which in this case was to familiarize students' journal reflection answers, while Step 2 is to create initial coding.
Step 3 is to categorize data into themes and subthemes.
Step 4 is to delete redundant themes and find codes. In this case, 15 codes from the post-lesson student reflection journals for online instruction, and 14 codes from the post-lesson student reflection journals for face-to-face instruction were developed.
Step 5 is to match themes that answer the research questions while Step 6 is to produce the findings. In this case, five themes emerged from the data relating to RQ2: (a) I-R-F sequence, (b) I-P-R-D sequence, (c) teacher-student interaction I-R-F & I-P-R-D, (d) intellectually cognitive processing I-R-F & I-P-R-D, and (e) learning challenges I-R-F & I-P-R-D. After the responses were elicited, iterative and inductive thematic coding were conducted by two experienced researchers following the previous established themes and the Cohen's Kappa coefficient for inter-rater coding was 0.89, showing a high level of coding reliability.
(3) Word-for-word transcription: The six-week spoken data audio-recorded between the instructor's questions and students' responses were typed out in a verbatim format to investigate both the number and content of the instructor's questioning and students' responses.

Students' Perceptions on EMI Content and Language Learning (RQ1)
Students' perceptions on teacher-student interaction for EMI content learning are presented in Table 2. Most of these participants (90.5%) indicated that interaction was important for learning. 80.9% of them expressed that EMI instructors should elicit substantial verbal responses from students rather than short answers. 35.7% of them disagreed that EMI instructors should do most of the talking in class, and 85.7% expressed that EMI instructors should know the students' existing understanding and their difficulties while teaching.  Table 3 demonstrates the students' perceptions on teacher-student interaction for English learning in EMI classes. 85.7% of these participants believed that their in-class English spoken output was necessary to improve their English, and 85.7% of them revealed that they needed regular opportunities to speak, and 92.9% of them expected lengthy spoken output. 85.7% of them expected their EMI instructors to correct their errors.

Differing Perceptions between the I-R-F and the I-P-R-D (RQ2)
As indicated in 2.6. Data Analyses, five overarching themes emerged after the hybrid process of inductive and deductive thematic coding analysis and are presented accordingly:

Perceptions on the I-R-F
50% of the participants (n=21) positively evaluated the instructor's use of the I-R-F sequence by using words such as improving concentration, interactive, attention-getting, helpful as presented in Table 4. The following excerpts and interpretations are elicited. S14: "Warm up with pictures can catch students' focus and it is easier to get involved and immersed into some abstract concept, such as culture, communication, and identity." (Improve concentration & fostering content knowledge) S8: "Helpful. Makes us quickly understand the structure of class and the content." (Usefulness) S26: "A good way to keep students focused. More interesting than traditional way when professors are talking to themselves only." (Improve concentration) S44: I got more opportunities to speak English." (Being interactive) though-provoking, interactive, and dynamic learning environment. Students' excerpts and interpretations are as follows.
S2: "Compared to I-R-F sequence, the instructor's I-P-R-D sequence helps me learn in a logical order….Also we're not just receiving information from him but trying to present our ideas and share with the whole class." (Logical and organized & being interactive) S10: "I really enjoy taking this course because the teacher's use of this I-P-R-D sequence is well-structured, and makes me understand abstract concepts." (Logical and organized) S21: "Delivered vivid and dynamic introduction and broad perspective, the way of explanation was clear and the conclusion helped us absorb the concept." (Dynamic learning environment) S42: "The sequence works well because it was stimulating and let us learn and understand the course content step by step." (Influential and thought-provoking & Logical and organized)

Teacher-student Interaction in I-R-F & I-P-R-D
24 students (57.1%) stated that the instructor's use of this I-R-F sequence had led to more of their participation as seen in Table 6. One excerpt is listed and interpreted below: S39: "I definitely had more participation because other EMI courses I had only had the output of professors." (Increased participation) 40 students (95.2%) stated that the instructor's use of the I-P-R-D had led to more of their participation as compared to the use of the I-R-F because various interactive learning activities implemented in class had inspired them to actively interact with the class and think more thoroughly. 2 students (4.8%) conveyed that they had less participation because they were not confident to express their ideas in public (see Table 7).  Table 8 demonstrates the salient differences regarding how students' perceptions on the I-R-F and the I-P-R-D differed. The I-P-R-D cyclic pedagogy outweighed the I-R-F sequence by 50% in terms of its effectiveness, by 19% in terms of leading to participation, and 45.2% in terms of helping students learn the content.

Intellectually Cognitive Processing in I-R-F & I-P-R-D
18 students (42.9%) approved that the instructor's use of the I-R-F sequence effectively facilitated their learning of the course content because they had more chances to think and speak as compared to other EMI online courses as presented in Table 9. More chances to think 5 5 40 students (95.2%) approved that the instructor's use of the I-P-R-D sequence effectively facilitated their learning of the course content because they had more in-depth comprehension of the course content and abstract concepts due to the well-structured and sequential pedagogy as compared to the I-R-F sequence of instruction (see Table 10). Based on students' Q4 responses in journal reflections, 14 students (33.3%) expressed that the I-P-R-D sequence of instruction was more effective than the I-R-F sequence to learn the course content because they were engaged in more active intellectually cognitive processing. Namely. They practiced analyzing (the 4 th level), evaluate (the 5 th level), and create (the 6 th level) of Bloom's taxonomy during the multimodal input of class activities. Some students' reflection excerpts are listed and evaluated as follows: S26: "I am more capable to analyze cross-cultural conflicts in the cases and give deeper explanations." (the 4 th level, analyze) S13: "The instructor gave us many cases that required us to compare some cross-cultural conflicts and behaviors." (the 4 th level, analyze) S4: "After an activity, the instructor often asked us what solutions we have. This is a good training because I force myself to give substantial solutions." (the 5 th level, evaluate) S38: "I enjoy the challenges the instructor gave us because I could think thoroughly about what an activity can be implied in our real lives." (the 5 th level, evaluate/synthesize) S20: "Many classmates have many ideas and this makes me to come up with more creative ideas so I could say something different." (the 6 th level, create) S17: "Participating in this class is not easy but I try to come up with my own ideas different from what everyone is saying and what everyone can think of." (the 6 th level, create) S1: "The use of this I-P-R-D sequence increased my participation effectively and encouraged me to think critically and present my own voice" (the 6 th level, create) 3.2.5 Learning Challenges in I-R-F & I-P-R-D Table 12 displays students' challenges to participate when the instructor used the I-R-F sequence as requested in journal Q5, 15 students (35.7%) expressed that this instruction required more attention and energy; 8 students (19.0%) recognized their lack of courage and confidence to speak English; 3 students (7.1%) stated their lack of critical thinking skills; 2 students (4.8%) expressed their limited English spoken ability; 1 student (2.4%) expressed that she couldn't interact in groups due to the constraint of an online environment, while 13 students (31%) indicated no apparent difficulties because there was more time to think before they spoke. Two excerpts and evaluated are shown below. S11: "I need to spend more energy and pay more attention on this class." (More attention and energy needed) S29: "I need to properly explain my thoughts and squeezing out an idea that hasn't been spoken yet, or it feels like I'm copying my classmate's idea." (More attention and energy needed) The challenges students faced in the I-P-R-D as inquired in journal Q5 included the need to think critically (n=17, 40.5%), the lack of courage to speak up for themselves (n=14, 33.3%), their inadequate vocabulary to express their thoughts in English (n=7, 16.7%), and the need to stay fully focused (n=1, 2.4%) as displayed in Table 13. The need to stay fully focused 1 1 No No apparent difficulty 9 9 4. Discussion

Students' Perceptions on Teacher-Student Interaction for EMI Content and Language Learning in Cross-cultural Communication Course
A sizable majority of these Taiwanese participants recognized the importance of teacher-student interaction for learning in an EMI course and were inclined to engage in extensive and substantial verbal output. This empirical evidence failed to tally An & Thomas' findings (2021) that most students recognized the merits of teacher-student interaction but were less inclined to engage in extensive interaction. Though both contexts (i.e., Taiwan & China) were under the influence of Confucian views that teacher-dominated teaching prevails, the discrepancy is probably due to the differences in study contexts and participants. The present study investigated an undergraduate EMI course that emphasized teacher-student interaction and pedagogy in Taiwan while the An & Thomas' (2021) study evaluated a more teacher-dominated learning context in EMI high-school settings in China.
The survey results for EMI language learning yielded an important finding that a relatively high percentage of the students (84%-91%) expected regular output chances, long verbal output, and corrective feedback from EMI instructors to improve their English language learning. Again, the results contradicted with An & Thomas' findings (2021) that both regular and lengthy output from students and error correction from teachers were not expected for most students in their study due to their limited English proficiency in engaging in interaction with the instructor.
Furthermore, half of the participants in the study accepted the instructor's monologue of lecturing; however, they recognized that teacher-student interaction remained important for learning the content of the course. This finding is interpreted that teacher-student interaction serves as a crucial contributor for instructors to know students' issues in learning the content. The finding also corresponds with the sociocultural constructivist's learning theory, verifying that knowledge is constructed, and misconception is changed through learning and interacting with others (Chin & Osborne, 2010).

Discrepancies between the I-R-F Classroom Discourse and the I-P-R-D Multimodal Input of Classroom Interaction
The I-P-R-D cyclic pedagogy had outperformed the I-R-F sequence by 50% in the effectiveness category, by 19% in the level of student engagement, and 45.2% in the effectiveness of helping students learn the content. This finding further supports why all of the students (n=42) agreed that the multimodal input of the interactive learning activities (e.g., the parable, tearing paper, making a paper chain, and critical incidents) in the I-P-R-D framework were not only fun, highly engaging, and thought-provoking but also helpful in engaging them to think more in depth and apply what they had just learned to the tasks. An alignment was noted that these students recognized that students' substantial, in-depth output was a key to high-quality teacher-student interaction (Littleton & Mercer, 2013).
By analyzing students' reflection excerpts, it was found that the multimodal input in the I-P-R-D sequence of elt.ccsenet.org English Language Teaching Vol. 16, No. 1; instruction had offered these students more opportunities to analyze scenarios, synthesize concepts, cooperate with others, evaluate appropriate solutions, discuss with one another, express their opinions, and create new thoughts and ideas. Namely, the I-P-R-D framework fostered learners' higher-order cognitive processing involving analysis, evaluation, and creation of new knowledge as well as more complex judgmental skills including analytical, critical thinking, and problem solving rather than simply learning, remembering, and applying facts or concepts as carried out in conventional I-R-F mode of classroom interaction. Thus, it is evident that the multimodal input of the I-P-R-D framework moved students' intellectual cognitive processing from their low-order learning outcomes to higher-order thinking skills as identified and presented in Bloom's revised taxonomy (Anderson & Krathwohl 2001) in Figure 5. The study results resonate with the claims that interaction-based pedagogies effectively facilitate learners' comprehension and instructors' lecturing communication (Flowerdew et. al., 2000;Kym & Kym, 2014;Suvinitty, 2010).

The limitations of the Study
The limitation of the study includes the limited items in questionnaires for evaluating students' beliefs toward EMI content and language. Moreover, the study results were not intended to be over-generalized because the EMI instructor in the study had nearly native-like English proficiency, which might have had an impact on students' experience and beliefs about classroom interaction, as studies indicated that EMI teachers' English proficiency was critical to engage in spontaneous speech in English and to discuss concepts in depth with students (Pun & Thomas, 2020; Zacharias, 2013).

Conclusion
This research probed 42 Taiwanese university students' perceptions on teacher-student interaction for EMI content and language learning in Cross-cultural Communication course, and how these students' perceptions differed between the I-R-F classroom discourse and the multimodal input of classroom interaction, the I-P-R-D.
The study yielded an important finding that the multimodal input of interaction not only increases students' engagement but also contributes to instructors' knowledge of students' issues in learning the content and fosters a richer, more in-depth experiential learning for the learners.