Computers & Education 182 (2022) 104483

Contents lists available at ScienceDirect

Computers & Education

journal homepage: www.elsevier.com/locate/compedu

Robots in situated learning classrooms with immediate feedback mechanisms to improve students’ learning performance

Vando Gusti Al Hakim a, Su-Hang Yang b,*, Mahesh Liyanawatta a, Jen-Hang Wang a, c, Gwo-Dong Chen a

a Department of Computer Science and Information Engineering, National Central University, No. 300, Jhongda Rd., Jhongli City, Taoyuan County, 32001, Taiwan

b Department of Hospitality Management, Chien Hsin University of Science and Technology, No. 229, Jianxing Rd., Jhongli City, Taoyuan County, 320312, Taiwan

c Research Center for Science and Technology for Learning, National Central University, No. 300, Jhongda Rd., Jhongli City, Taoyuan County, 32001, Taiwan

A R T I C L E I N F O

Keywords:

Augmented and virtual reality Human-computer interface Teaching/learning strategies Improving classroom teaching Cooperative/collaborative learning

1. Introduction

A B S T R A C T

Because of time and space restrictions and the limited interaction capabilities of robots, it is preferable for teachers to construct a learning environment using digital reality in conjunction with robots. Doing so enables students to learn and interact in any scenario relevant to textbook material that can be effectively digitalized, while also promoting human-robot interaction. Here, an interactive situated learning approach was developed to improve students’ learning perfor-mances. The students and robot role-played characters and immersed themselves in digital sit-uated learning tasks and challenges. This approach included a real-time feedback mechanism to guide and evaluate the knowledge application of the students. The evaluation was performed during interactions with the robot, virtual objects, and virtual characters based on textbook context and content. The experiment was conducted during an English as a second language course for junior high school students. A total of 101 students were assigned to three groups with different approaches and their learning performance was evaluated. The experimental results indicated that students who learned with the proposed approach exhibited better learning achievement and significant positive effects in terms of learning motivation and engagement. Furthermore, interaction with physical robots improved student learning achievements signifi-cantly compared with virtual interaction. Moreover, motivation in the learning process could be enhanced using authentic objects and scenarios in the digital situated learning environment.

Recently, the potential of using robots has been explored in the field of education, and several studies have demonstrated that robots can play an important role in learning (Belpaeme & Tanaka, 2021). In a classroom environment, robots can offer benefits as teaching tools that can be matched with specific instructional goals based on their characteristics, such as repeatability, flexibility, the ability to present digital data (digitization), interactivity, appearance, and body movement, to support learning (Chang, Lee, Chao,

* Corresponding author.

E-mail addresses: [email protected] (V.G. Al Hakim), [email protected] (S.- H. Yang), [email protected] (M. Liyanawatta), [email protected] (J.- H. Wang), [email protected] (G.- D. Chen).

https://doi.org/10.1016/j.compedu.2022.104483

Received 16 October 2021; Received in revised form 17 January 2022; Accepted 21 February 2022

Available online 25 February 2022

0360-1315/© 2022 Elsevier Ltd. All rights reserved.

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Wang, & Chen, 2010). They also increase curiosity in students (Gordon, Breazeal, & Engel, 2015) and render interactions more enjoyable and engaging (Saerbeck, Schut, Bartneck, & Janse, 2010). Moreover, the physical appearance of a robot allows student learning through embodied interactions using its sensors, such as touch (Barnes, FakhrHosseini, Vasey, Park, & Jeon, 2020), dialogs (Ahmad, Mubin, Shahid, & Orlando, 2019; Babel et al., 2021), and movements or gestures (Kanda, Hirano, Eaton, & Ishiguro, 2004; Kose¨ et al., 2015). Robots are becoming increasingly relevant as educational tools, illustrating their significance. In general, while conducting robot-supported instructional activities inside the classroom, the teachers insert the learning material into the robots, which display this material through their integrated screens (e.g., Chen, Park, & Breazeal, 2020; Gordon et al., 2015; Jones & Cas-tellano, 2018; and Kennedy, Baxter, Senft, & Belpaeme, 2016). Thus, the students can learn the material through the screen while socially interacting with the robots.

Conversely, according to the theory of situated cognition postulated by Brown, Collins, and Duguid (1989), learning should occur in context and obtained knowledge should be applied in real-life practices. Generating a learning environment with learning contexts has become an option to support situated learning activities. In such a learning environment, the students can reflect on the application of knowledge and skills in real-life scenarios (Bosman & Fernhaber, 2019). To understand how students apply their knowledge, teachers typically select scenarios based on textbook content and context. In the classroom environment, robots often serve as a physical pedagogical agent for mediating interaction between students and tangible objects, allowing the students to immerse themselves into the situational learning environment for practice, learning, and interacting with objects as well as each other. Consequently, the teachers or publishers may need to build real-life sets and props using tangible objects and toys to emulate the situational learning environment (Cheng, Wang, Yang, Yang, & Chen, 2020). However, a textbook may contain many different contexts in each chapter, making it difficult for teachers to provide sufficient context and scenarios to represent every situation dis-cussed in the textbook for students to learn inside the classroom (Chen, Fan, Chang, Chang, & Chen, 2018; Hwang & Wang, 2016). Anderson, Reder, and Simon (1996) indicated that building proper scenarios and real-life props or objects for situated learning in a classroom is challenging; whereas, constructing a learning environment using digital reality to allow both robots and students to learn and interact in the scenario required for the learning content is easier.

Owing to the gradual development of digital reality (e.g., 2D/3D scenarios, augmented reality, virtual reality, and mixed reality), real-life scenarios and any situation corresponding to learning activities can be easily created in a classroom. With the aid of a motion-sensing device, such as Kinect, images and actions of characters can be integrated with built-in digital scenarios and projected onto a screen. This allows the use of virtual props and interaction between characters (Zhang, 2012). Kinect enables the quick and appropriate set up of digital situational learning environment in regular classrooms. Moreover, during situated learning activities, the characters may encounter problems and require a companion to correct and guide them to complete their tasks or missions. This companion may provide guidance or hints situationally to help students learn through social interaction (Cheng et al., 2020; Okita, Ng-Thow-Hing, & Sarvadevabhatla, 2009) and a direct status and instant feedback, leading the students to inquire and obtain solutions to the problems encountered (Barnes et al., 2020; Bravo Sanchez,´ Gonzalez´ Correal, & Guerrero, 2017). However, there is no learning environment suitable for regular classrooms that quickly transforms learning content and contexts from textbooks into digital scenarios and allows students to immerse and interact in a situated learning activity with immediate feedback to guide and evaluate them.

This study aims to fill the gap by creating an interactive situational learning environment with a robot as a companion to perform and provide situational real-time evaluation and guidance during digital situated learning tasks. The aforementioned environment allows students to learn in an authentic virtual world with immediate evaluation to determine whether they could apply the existing knowledge by interacting with the robot and virtual objects and characters. The combination of technologies and sensors of such a learning environment can provide embodied activities that offer meaningful, multimodal, and playful experiences for learners inside the classroom (Kosmas, Ioannou, & Zaphiris, 2019). A rich sensory environment combined with an interactive learning context may positively impact student engagement (Barnes et al., 2020; Chin, Hong, & Chen, 2014). Teachers can design scripts in which the robot plays a guide and evaluator for students while they attempt to obtain a digital object or pass a gate into the next stage of the scenario. Consequently, the students must focus on learning materials embedded in the scenarios, which may promote their motivation to learn and compel them to study beforehand to perform well in the situated learning activity. Promoting learning motivation has been recognized as an important issue because it is a critical index for evaluating learning achievement (Legault, Green-Demers, & Pelletier, 2006; McCombs, 2000). Several studies have discussed this issue because robust learning motivation can improve learning, enhancing learning performance (Murphy & Alexander, 2000; Viljaranta, Lerkkanen, Poikkeus, Aunola, & Nurmi, 2009). However, it is chal-lenging to design a learning approach that improves learning performance in terms of learning achievement, motivation, and engagement.

This study proposes an interactive situated learning approach to investigate the learning performances of students. Three learning approaches are compared, and their effects are assessed. The first approach is interactive situated learning, wherein a group of students and a robot are immersed in a virtual world where they interact with scenarios, virtual objects, virtual characters, and the robot’s recognition capabilities (i.e., physical touch, embodied gestures, and dialogs). The second approach involves a situated learning environment as in the first approach but without the robot’s recognition capabilities. The last approach is conventional instruction, wherein a group of students performs with a robot on a traditional dais in front of the classroom without being immersed in a virtual world, and the learning material is displayed on the robot’s screen. Furthermore, an experiment is conducted during an English as a second language course for junior high school students by employing robots with different learning approaches to investigate the following questions.

1. Do students who learn with the interactive situated learning approach show better learning achievements than those who learn with a situated learning approach or conventional instruction?

2

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

2. Do students who learn with the interactive situated learning approach show higher learning motivation than those who learn with a situated learning approach or conventional instruction?
3. Do students who learn with the interactive situated learning approach show higher learning engagement than those who learn with a situated learning approach or conventional instruction?

2. Literature review

2.1. Situated learning

Situated learning was proposed by Lave and Wenger (1991) as a learning model in communities of practice; they argued that learning was a social process that co-constructed the information situated within a specific context. Situated learning might occur inside the classroom by placing learners in certain situations. Providing an authentic environment was critical to applying situational learning in the classroom (Herrington & Oliver, 1995). This environment aided in placing learners in various situations and conditions, such as those they experience outside of the classroom in their daily lives. There should be a theorized relationship, supported by a range of digital technologies, between the designed environment and the situated learning activity to accommodate teachers or publishers (Yeoman & Wilson, 2019).

In situated learning contexts, teachers could design a learning environment with authentic contexts that allowed students to practice and implement their knowledge. Students were expected to integrate their previously acquired experience and knowledge with recently obtained knowledge to apply it in the real world. They should learn inside the environment using their cognition to construct knowledge and experience meaningful learning (Sung, Hwang, & Yen, 2015). Thus, a situated learning environment aimed to develop the ability to implement the knowledge acquired from school in real-world contexts. Providing learning contexts or ma­ terials for the students to experience what they might encounter in their daily lives has become essential.

Over the past few years, numerous scholars have adopted the situated learning approach to improve students’ likelihood of solving authentic problems in real practice in fields ranging from mathematics (Anderson et al., 1996) to natural disasters (Chou et al., 2012). Collentine and Freed (2004) claimed that placing students in authentic situational contexts was crucial for learning languages. Further, a study conducted by Chen and Lin (2016) corroborated this claim. They proposed a digital game-based situated learning approach to teaching Chinese poetry. The group of students that used this game outperformed the group that received traditional instruction, demonstrating that situated learning positively affected learning. Also, Hwang and Wang (2016) presented a situated role-playing computer game to engage students in the corresponding English vocabulary contexts with different test-item guiding approaches. The game provided vocabulary flashcards showing the pronunciation of a word, cloze and multiple-choice guiding strategies, and hints to help students complete tasks in the game. Their findings suggested the importance of providing diverse learning tasks with chal­ lenges that the students could overcome and that could assist them in the learning process.

In contrast, earlier studies combined situated learning with educational drama to build digital situated learning for students to perform collaboratively in the classroom. For example, Chen et al. (2018) proposed a situational learning system involving a comic play and scriptwriting. They claimed that by learning through dramatized stories along with immersion provided by the system, students could enhance their knowledge application and acquire autonomy and ownership of the learning process and strong im­ pressions of the creative work. Furthermore, Liyanawatta, Yang, Liu, Zhuang, and Chen (2021) introduced a methodology to transform a classroom into a digital learning theater, allowing performers and audiences to participate and engage in performances collabo­ ratively. Unlike conventional theater, the digital learning theater allowed actors to observe, encounter, and perform the scene on a virtual stage, which was similar to a mirror room, while their classmates, the audience, watched them perform via the screen at the front of the classroom. The digital learning theater, which included an approach with tablet personal computers, enhanced student learning effectiveness through performing, watching, interacting, and participating in the learning context.

Using digital reality for situated learning allowed teachers to quickly build scenarios, prepare props, and guide students in situ­ ational learning environments inside the classroom based on textbook content and contexts. Kinect enabled quick building of such a learning environment and facilitation of collaborative learning for up to six people in regular classrooms. Furthermore, students used their embodied gestures and interacted with virtual objects, virtual characters, and props to further enhance their learning experiences in a digital situational learning environment. However, during the performance, talented actors were required to generate an apparent affect and convey the situations to engage the audiences (Murphy et al., 2011). Students were generally not professional performers; therefore, the engagement of the student actors in performing and the interest of the audience might diminish if the learning activities were not engaging in terms of situational interaction. Furthermore, students might not be sufficiently confident and able to reflect and correct errors in their performances immediately, leading to a poor performance. Thus, there should be a feedback mechanism in the situated learning environment to guide and evaluate students during the performance.

2.2. Learning environment with robots

The advancements in research on robotics enabled robots to assist humans in many ways. In the context of education research, several researchers believe that robots exhibited great potential to improve learning effectiveness. For instance, an interactive hu­ manoid robot in Kanda et al. (2004) could grasp the attention of the learners and assisted them in developing meaningful interactions. Moreover, robots have also been shown to improve learning interest (Chin et al., 2014), enhance learning motivation (Chang, Lee, Chao, et al., 2010), and encourage self-regulated learning in students (Jones & Castellano, 2018). Chen et al. (2020) stated that social robots held great potential to serve as emotionally appealing learning companions for learners in social contexts. Similarly, several

3

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

scholars believed that more humanlike robot features could improve students’ learning performance (Okita et al., 2009), acceptance, and trust (Babel et al., 2021). However, Kennedy et al. (2016) and Konijn and Hoorn (2020) presented a contrasting result, where additional efforts in developing robot behavior did not have a positive impact. Researchers in the field continue to debate the aspects of robot behavior design that support learning (Saerbeck et al., 2010).

There were various robots for learning that perform different roles. Chang, Lee, Chao, et al. (2010) identified the roles of educational robots in the learning environment, including learning materials, pets or learning companions, and teaching assistants. They found that robots had potential roles as instructional tools to support teachers, provide an interactive and engaging learning experience, and effectively promote learning motivation. Belpaeme, Kennedy, Ramachandran, Scassellati, and Tanaka (2018) con-ducted a meta-analysis of the literature on robots for learning. Their study determined that aside from role-playing as a tutor or a teacher, robots could also be learning companions as well as novices to allow students to take on the role of an instructor. Furthermore, they found that robots as learning companions in peer-to-peer interactions were less intimidating than when tutoring or teaching, providing significant benefits.

Regarding activities in a classroom environment accompanied with robots, Cheng, Wang, and Chen (2019) provided a design framework to guide researchers or teachers to construct an immersive language learning environment. In this environment, children played with Internet of Things toys and were assisted by a robot, which acted as a companion and socially interacted with students through parent-like linguistic feedback in the target language. Also, Chang, Lee, Wang, and Chen (2010) presented a RoboStage system by importing robots into an authentic mixed-reality environment with tasks for educational purposes. They reported that students preferred the tangible user interface and physical robot interactions over virtual ones. The study claimed that students found this strategy enjoyable and engaging, which affected their learning motivation. Sugimoto (2011) proposed a system GENTORO that employed a robot and handheld projector to support storytelling for children in mixed reality. The system promoted children’s learning motivation and engagement and assisted them in presenting their achievements. Antle and Wise (2013) stated that a hybrid phys-ical–digital user interface could benefit learning owing to the unique and novel interactions provided by augmented environments and the embedded controls of the system.

Considering the benefits of an authentic learning environment and the role of a learning companion, this study presents a digital situated learning environment with a robot to allow both students and the robot to immerse and perform in the virtual world. The performance was constrained by learning script based on the structure and roles in Christopher Vogler’s “The Writer’s Journey” (2007). For the learning activities to be meaningful, the students interacted with a physical robot and virtual objects and characters through embodied interactions based on textbook content and contexts while performing in the learning environment. The robot was assigned character archetypes to guide and evaluate students by recognizing their interactions, such as touch, gesture, and dialog during their performance according to the script. Therefore, the learning environment of this study offered a situational real-time feedback mechanism to communicate whether students performed correctly according to the learning content and to enable them to decide how to proceed within the learning script.

3. Development of the interactive situated learning approach

3.1. Interactive situated learning model

Here, an interactive situated learning approach was developed by employing a robot aimed at embodied recognition as a com-panion inside a virtual world to achieve situated learning activity in the classroom. The learning approach was developed by referring to the tangible learning design framework proposed by Antle and Wise (2013) and considering tangible user interface features, in-teractions, and learning (see Fig. 1 left). The taxonomy and guidelines generated from the cognitive and learning theories in the tangible learning design framework were used as references for designing the learning approach here. The robot and students were

Fig. 1. Developed learning approach (left) and environmental setting (right).

4

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

immersed in the virtual world by standing in front of Kinect v2, and they interacted with scenarios, virtual objects, virtual characters, and robot. The modalities of interactions to engage students in both the physical and virtual worlds were the recognitions of the robot based on physical touch, gesture, and dialog. In the virtual world, a scenario and its corresponding task on the computer screen were presented to guide students in speaking, moving, and acting according to the story narrative. Brown et al. (1989) stated that a meaningful experience can only occur if it is embedded within the social and physical context in which it will be applied. To support meaningful content-related goals, the activities occurred in an authentic scenario with interactive virtual elements, such as dynamic background, virtual objects, and virtual humans. Students could therefore enter and experience everyday situations and apply their knowledge through interactions in the virtual world with scenarios, the robot, and virtual objects and characters. Students who perform in the virtual world are directly involved in solving the tasks or challenges, which necessitates responsibility and knowledge application to complete the journey.

Here, the classroom setting was separated into a performers’ area and audience area (see Fig. 1 right). A group of students (hereafter, “student actors”) and robot performed in the performers’ area and watched their performance in a virtual world similar to a mirror room from the computer screen. The robot and virtual world provided real-time situational feedback, which was displayed on the computer and robot screens. These included guidance and evaluation to ensure that students acquired the knowledge; therefore, learning activities became interactive. Owing to the real-time situational guidance and evaluation, student actors could reflect on themselves and mutually adapt while performing. The reflexive processes during learning activities allowed students to realize their actions, understand the reasons behind them, and determine new actions. Also, these processes enhanced motivation to help students amend their actions (George, Michel, & Ollagnier-Beldame, 2016). The remaining students sat in the audience area equipped in front with a screen projector, connected to a computer, so they could simultaneously watch, monitor, and learn from the performances of the student actors. Here, the audience are positioned as watchers/spectators without any involvement in the virtual world.

3.2. Structure of the interactive situated learning system

Fig. 2 illustrates the system architecture. The controller for the virtual world, developed on the Android platform, was installed on teachers’ tablets. Moreover, the display interface of the virtual world program was constructed using the C# programming language. The virtual world program was installed on a personal computer with an NVIDIA graphics card and running on Windows 10. Microsoft Kinect v2 SDK attached to a personal computer was used to capture and recognize the skeletons and images of the actors. The per-formance was video recorded through the built-in Xbox Game Bar in Windows 10. The system database employed a webserver that used WampServer v2.4.9 as the platform and MySQL v5.6.17 as the database environment to store all the information of the learning system.

This study employed a commercial robot, ZENBO, from ASUS. The Android-based robot had one screen to display a face with changing expressions. It was 62 cm tall with a round body, a touch sensor on its head, and two wheels with various LED colors to support mobility. The robot included several features, such as display of emotions (e.g., facial expressions, wheel lights, action), multimedia deployment (e.g., displaying images and playing sound effects), and Wi-Fi connectivity to support integrating third-party programs to extend its capabilities. Microsoft Azure Speech Cognitive Services and DialogFlow were employed to manage the robot dialog capability. Furthermore, Kinect v2 was used to enable the robot to recognize gestures. All the system components (i.e., robot, tablet control, virtual world program, database) employed socket communication. The combination of the aforementioned system components can generate gesture, touch, and dialog recognition, thus enhancing the interaction capabilities of robots. Moreover, the robot’s feedback is also made more interactive through sound effects, the virtual scoring system, and the dynamic scenario(s) accessed through virtual world.

During the situated learning performances, the teacher situationally controlled the flow of the scenarios, robot, and interactions or recognitions based on the script through the Performing System Interface on the tablet. Furthermore, this interface managed which subtitles were displayed on the screen to be read by the humans, robot, or computer (see Fig. 3 left). When the teacher clicked on the subtitles, the subtitles were displayed in the virtual world (see Fig. 3 right). Consequently, prior to the performance, the teacher had to

Fig. 2. System architecture.

5

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

select the interaction modes and virtual elements on the Authoring System Interface on a tablet to write plots, construct the virtual world scenario, and input learning materials quickly based on textbook content and contexts. The learning materials encountered in the form of challenges (e.g., keywords, actions, and gestures) in a particular scenario had to be learned by students to complete their journey.

The learning system captured images of the performers during their performance on the stage and combined them into the virtual world. The system could detect the gestures and dialogs of the performers as well as the touches made to control interactive objects. Student actors could focus on their acting and interact with the robot, scenes, and virtual props to complete their performance. The performers’ acting merged with the digital scenario on the performers’ area was projected on the big screen in the classroom and simultaneously recorded. Thus, after a class, students reviewed and reflected on their performances.

3.3. Storyline and interactions in the interactive situated learning system

Here, learning material from an English course textbook was compiled into a learning script. The script was developed in collaboration with schoolteachers who have taught the English course for over a decade. The designed script positioned the robot as mentors, messengers, or threshold guardians, and the student actors had to obtain information, acquire virtual objects, or pass a gate to proceed to the next stage of their journey. By constraining the performances and tasks with a learning script, this strategy can solve action coordination (Hoffman, 2011) and role sharing (Lemaignan, Gharbi, Mainprice, Herrb, & Alami, 2012) between humans and the robot in the same stages. All experimental and control groups used the same learning script. To provide the narrative structures and character development present in most stories, the script was written based on Writer’s Journey: Mythic Structure for Writers by Christopher Vogler (2007), and Transformational Play by Barab, Gresalfi, and Ingram-Goble (2010) was used for positioning actors, context, and content.

In short, the entire script described a journey of the protagonists, heroes, to retrieve the president caught by a Coronavirus devil in a ghost town. The student actors represented heroes with specific characteristics, who were willing to sacrifice themselves, and they made decisions to complete the mission. The other character was a mentor who taught and protected the heroes from opponents and gave them gifts. The robot was designated as a mentor archetype, and owing to its multimodal functionality, it could manage both the physical and virtual environments. The heroes could obtain the virtual objects provided by the virtual world through physical manipulation provided by robot. Namely, the robot could fulfill various needs of the heroes to continue their journey. Moreover, the robot also had roles in the script as a messenger that guided the students and a threshold guardian that evaluated them. Thus, the students had to focus and use the learning materials from the learning script to complete a task or proceed to the next stage of the journey. The last character was a shadow who represented the dark side as an opponent or enemy of the heroes. The shadow here was a virtual character provided by the virtual world.

Fig. 4 depicts the storyline with a detailed description of the interactions. Each group of student actors had to pass all nine stages of the journey to complete it. Each stage had a corresponding scenario, characters, subtitles, and story adhering to the screenwriting principles of Vogler (2007) and the transformational play theory of Barab et al. (2010). The virtual world provided an interactive score, virtual objects, and scenarios that adaptively changed with the situations. The entire journey consisted of four robot interactions that the student actors had to complete through producing an appropriate action, gesture, or dialog to obtain an object or pass a gate to move to the next stage of the journey.

The first robot interaction comprised a scenario wherein a group of heroes asked the robot to take them to the ghost town. To obtain a pass to the ghost town scenario, heroes had to correctly answer the question posed by the robot (see Fig. 5[1]). The robot activated

Fig. 3. Tablet control (left) and virtual world (right).

6

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Fig. 4. Storyline of the proposed approach.

dialog recognition for this interaction. Fig. 5(2–3) depict the second robot interaction comprising a physical touch to the head of the robot to obtain a digital mask for protecting heroes from coronavirus. Fig. 5(4–5) show the third robot interaction involved attaching or wearing the obtained mask on the heroes’ faces by performing real tasks, i.e., the gesture of covering their mouths with their hands. During this interaction, the system activated gesture recognition. After the heroes put on the digital masks and were ready to defeat the enemy, they were to face the greatest challenge and to experience death and rebirth in The Ordeal stage. In this stage the robot gives them the wrong weapon, thus making the shadow/enemy able to dodge the attack and produce an evil laugh. The fourth robot interaction was identical to the second one; however, the objective here was to defeat the enemy (see Fig. 5[6]). Every feedback from the robot and the virtual world enabled mutual adaptation based on the responses of the student actors.

Fig. 6(1) shows that when the students correctly performed the indicated actions, gestures, or dialogs, the robot provided positive, happy feedback and praised them. The virtual world then changed the scenario, increased the score, and/or attached the virtual objects to the bodies of the student actors (see Fig. 6[2]). Conversely, when student actors failed the interaction, the robot provided unhappy feedback (see Fig. 6[3]), and a few hints and motivational messages. Hints and motivational messages were provided to guide the actors to replay the interactions and to encourage them to contemplate and obtain the correct answers. The motivational messages were based on Rajendran, Iyer, and Murthy (2018) who claimed that such messages could decrease the frustration levels of students

7

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Fig. 5. Illustrative examples of the virtual world and robot interactions.

Fig. 6. Illustrative examples of the system feedback.

8

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

and help them continue their session. In this case, the virtual world decreased the score and attached the virtual burned masks to the faces of the student actors (see Fig. 6[4]).

4. Methodology

A quasi-experiment was conducted here to evaluate the effectiveness of the proposed learning approach in terms of learning achievement, motivation, and engagement among three groups (i.e., experimental Groups A and B and the control group). Each group underwent different modes of learning with robot. The measurement tools used here were: (1) a pre- and post-test to examine the students’ learning achievement, and (2) questionnaires to understand their experience regarding learning motivation and engagement.

4.1. Participants

The experiment was conducted in an English as a second language class at a junior high school in Taiwan. It occurred during the summer break of 2020, immediately following the completion of the students’ final examinations. A total of 101 ninth graders from three classes were recruited to participate in this experiment. Three classes were randomly selected as the experimental Groups A and B and the control group. Experimental Group A consisted of 39 students (14 females and 25 males) and received the interactive situated learning approach. Group B comprised 31 students (13 females and 18 males) and learned with the situated learning approach. Finally, 31 students in the control group (11 females and 20 males) completed activities using conventional approaches to instruction. All learning approaches used one robot and had the same classroom environmental setting, i.e., a performers’ area and an audience area. All students had the same opportunities to take turns to perform as student actors. All subjects were 15 years old and taught by the same English teacher. The different levels of achievement of the students in class resulted in creating heterogeneous groups. The class teachers selected the students with high achievement levels as group leaders; each group had at least one group leader. All three groups used a similar learning script. The three classes were completely independent during the experiment and did not affect each other.

4.2. Measurement tools

The measurement tools here included a pre- and post-test to examine students’ learning achievement, as well as questionnaire to measure the students’ learning motivation and engagement.

The pre- and post-test questions consisted of five fill-in-the-blank and five short-answer items with scores ranging from 0 to 100. The questions were written by three English teachers with over a decade of experience in teaching English. The questions were designed to measure learning achievement in terms of vocabulary, grammar, and sentence patterns.

The learning motivation questionnaire proposed by Hwang, Yang, and Wang (2013) was modified to develop the learning moti-vation measure for this study. It consisted of seven items with a six-point Likert scale format, such as “I will practice the script diligently when participating in the journey scenario together with the robot.” The reported reliability coefficient (Cronbach alpha) of the scale is 0.79 (N = 56).

The learning engagement questionnaire was adopted based on the measure of cognitive engagement presented by Jamaludin and Osman (2014). It consisted of four items, such as “Interacting with the robot on a journey can deepen my understanding of the learning content,” with a seven-point Likert rating format. The Cronbach alpha value obtained in the previous study was 0.955 (N = 24), showing good reliability.

Owing to the low English proficiency of Taiwanese students, all the questions in the questionnaire were translated into Mandarin and validated by experienced researchers and teachers in the school before their implementation in the experiment.

4.3. Procedure

Fig. 7 shows the experimental procedure, which consisted of six stages: (1) the pre-test, (2) system demonstration, (3) learning

Fig. 7. Experimental procedure.

9

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

script and rehearsal, (4) the group treatments, (5) post-test and questionnaire, and (6) interview.

In the first stage, all the students from all three groups completed the pre-test in 10 min. The pre-test aimed to evaluate students’ learning achievements prior to interacting with our system. After completing the pre-test, and still in the first week, all students moved on to the second stage, i.e., system demonstration. This stage was included to help students understand and become familiar with the learning environment, process, and tasks. It included a tutorial on how to interact with the system. Over the second and third weeks, the learning script and rehearsal were provided to familiarize the students with the script and performing system. In this stage, a schoolteacher was also involved in the experimental process to teach students the content of the textbook (i.e., vocabulary, grammar, and sentence pattern) and explain the meaning of each sentence in the script. Then, the teacher gave instructions to the students regarding learning materials from the script, assisted students in practicing the script (e.g., gestures or actions), and helped them learn pronunciation correctly. Subsequently, the students were divided into smaller groups (five to six students in each group) to take turns rehearsing using the system. During this stage, students were allowed to bring their scripts to practice performing after class. The total time required for the learning script and rehearsal stage was 90 min.

In the fourth stage, each actor group delivered a formal performance using the system while their classmates, the audience, could reflect and learn by watching them. Experimental Group A delivered a formal performance with a robot in the interactive situated learning environment. The students in Group B performed with the robot in the situated learning environment. Subsequently, the students in the control group delivered a formal performance with a robot on a traditional dais. After each group completed its performance, the teacher discussed the performance and permitted the audience to advise the performers. In the fifth stage, all students were asked to complete the post-test and complete the questionnaires in 10 min, respectively. The post-test questions were the same as the ones in the pre-test given in the first stage to maintain consistency in the test results. The English teacher of the classes graded the pre- and post-test answers. Here, a single factor analysis of covariance (ANCOVA) was employed to examine the learning achievements among the three groups based on the obtained test scores (i.e., pre- and post-tests). The questionnaires to better understand the perceptions of the students regarding “motivation” and “engagement” among the three groups after the treatment were analyzed with an analysis of variance (ANOVA). In the final stage, ten students were selected from each group to be interviewed to determine their opinions about the learning approach.

5. Results

5.1. Learning achievement analysis

Here, there were 101 valid pre- and post-test samples, including 39 in Group A, 31 in Group B, and 31 in the control group. The experimental data were analyzed by an ANCOVA, using pre-test scores as the covariate, post-test scores as the dependent variables, and the different learning modes as the independent variable. The ANCOVA was employed to compare the post-test scores of learning achievement by excluding the pre-test score impacts. It was used to examine the effectiveness of the interactive situated learning approach in enhancing students’ learning achievement.

Before conducting the ANCOVA, an ANOVA was conducted to verify whether the three groups had similar prior knowledge. As Table 1 shows, there was no significant difference in the pre-tests of the three groups (F = 0.367, p > .05); this suggested that the three groups had equivalent knowledge prior to the experiment.

The assumption of the homogeneity of regression slopes was not violated (F = 1.976, p > .05), meaning that it was reasonable to conduct ANCOVA. Table 2 presents the ANCOVA results. After excluding the pre-test score impact, a significant difference between the three groups was found (F = 11.58, p < .05), implying that the three groups had significant differences in the post-test scores. Using pairwise comparisons, it was found that experimental Group A outperformed experimental Group B and the control group, demon-strating that learning with a robot in an interactive situated learning environment could significantly improve students’ learning achievement, compared to a non-interactive model one and the conventional instruction model. Moreover, it was also found that the control group had significantly higher learning achievements than Group B, implying that learning with a robot in the traditional dais in front of the classroom has better effects on learning outcomes of students than the situated learning approach (in which the robot and student actors were immersed into a virtual world, but have no robot recognition [non-interactive]).

Additionally, to identify what dimensions of learning achievement benefited from the proposed learning approach, a series of ANCOVAs was conducted with the pre-test scores of each dimension of learning achievement as the covariate, the different learning modes as the independent variable, and the post-test scores of each dimension of learning achievement as the dependent variables.

The assumption of the homogeneity of regression slopes was not violated as demonstrated by the following results: vocabulary, F = 1.555, p > .05; grammar, F = 1.405, p > .05; and sentence patterns, F = 0.365, p > .05. It was therefore reasonable to conduct ANCOVA to identify the differences between the three groups in the three dimensions.

As Table 3 shows, a significant difference in all three dimensions between the three groups was found, indicating that the three

Table 1

ANOVA results on the pre-test scores of the three groups.

	Group	N	Mean	SD	F
EA	Experimental Group A	39	46.34	15.58	.367
EB	Experimental Group B	31	48.62	11.73
C	Control group	31	45.45	17.17

10

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Table 2

ANCOVA results of student learning achievement.

	Group		N	Mean	SD		Adjusted mean	F (2, 97)	Pairwise comparisons
EA	Experimental Group A		39	79.18	18.14		79.54	11.58*	EA > EB
EB	Experimental Group B		31	66.05	19.44		64.45		EA > C
C	Control group		31	71.36	17.62		72.51		C > EB
	Total		101	72.75	19.03
*p < .05.
Table 3
ANCOVA results of student learning achievement in the three dimensions.
Dimension			Group	N	Mean	SD	Adjusted mean	F (2, 97)	Pairwise comparisons
Vocabulary		EA	Experimental A	39	39.37	9.37	39.62	11.47*	EA > EB
		EB	Experimental B	31	32.97	9.66	32.13		EA > C
Grammar		C	Control	31	35.57	9.38	36.10	9.49*	C > EB
		EA	Experimental A	39	26.66	7.40	27.33
		EB	Experimental B	31	23.49	7.56	22.27
Sentence patterns		C	Control	31	24.30	6.78	24.67	7.24*
		EA	Experimental A	39	13.14	6.62	12.47
		EB	Experimental B	31	9.59	5.76	9.99
		C	Control	31	11.49	5.66	11.94

*p < .05.

groups had significant differences in their post-test scores. Furthermore, the pairwise comparison results were consistent with previous ANCOVA results. In summary, the results above demonstrate that, compared to a non-interactive approach and the conventional instruction approach, the students who learned with a robot in an interactive situated learning environment approach exhibited greater achievement in the dimensions of vocabulary, grammar, and sentence patterns.

5.2. Learning motivation analysis

The responses to the questionnaires were collected to understand the students’ perceptions of learning motivation after partici-pating in the different learning modes with the robot. A total of 100 out of 101 completed questionnaires were included as valid copies. Table 4 shows the ANOVA results of students’ learning motivation. It was found that the three groups had significant differences in learning motivation (F = 23.92, p < .05). The pairwise comparisons showed that experimental Group A outperformed both experi-mental Group B and the control group, implying that the interactive situated learning model outperformed the other two models in terms of learning motivation. Moreover, it was found that experimental Group B outperformed the control group, indicating that both situated learning approaches with interactive and non-interactive modes showed significantly higher learning motivation in com-parison with the conventional instruction approach.

5.3. Learning engagement analysis

A total of 100 out of 101 learning engagement questionnaire responses were selected as valid copies. Table 5 lists the experimental results. The obtained F-value (F = 22.61, p < .05) indicated significant differences between the learning engagement in the three groups after participating in the different modes of learning. Based on the presented pairwise comparisons, it was found that exper-imental Group A outperformed experimental Group B and the control group. However, in terms of learning engagement, there was no significant difference between experimental Group B and the control group. The result indicated that learning with a robot inside an interactive situated learning environment showed better learning engagement than a non-interactive model and conventional in-struction model.

Table 4

ANOVA results of students’ learning motivation.

	Group	N	Mean	SD	F	Pairwise comparisons
EA	Experimental Group A	39	4.05	.72	23.92*	EA > EB
EB	Experimental Group B	30	3.33	.71		EA > C
C	Control group	31	2.97	.55		EB > C
	Total	100	3.50	.81

*p < .05.

11

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Table 5

ANOVA results of students’ learning engagement.

	Group	N	Mean	SD	F	Pairwise comparisons
EA	Experimental Group A	39	4.05	.79	22.61*	EA > EB
EB	Experimental Group B	30	2.77	1.10		EA > C
C	Control group	31	2.90	.74
	Total	100	3.31	1.06

*p < .05.

5.4. Interviews with students in Experimental Group A

To obtain the students’ opinions regarding the proposed learning approach, 10 students in experimental Group A were randomly selected for interviews. The interview questions were obtained from a questionnaire designed by Chang, Lee, Wang, and Chen (2010). These questions were modified to correspond to the effectiveness of learning with a robot in an interactive situated learning envi-ronment, for instance, “Which part of the learning activities made you feel good about your learning? Please give an example to illustrate your point.” The purpose of the interview was to understand students’ perceptions and to investigate the possible factors that affect their willingness to learn in such a learning approach, with the aim of improving our learning system in the future.

Based on the results of the interview with the students in experimental Group A, it could be inferred that “compelling,” “inter-esting,” and “fostering understanding” were the notable features of learning with a robot in an interactive situated learning environment.

In terms of “compelling,” six students (A2, A4, A5, A7, A8, and A10) shared similar opinions that learning with a robot in an interactive situated learning environment promoted their learning motivation. The students also believed that this strategy could compel them to study beforehand by practicing after class with a script to perform well during their performance. For instance, A5 revealed that “Because the learning materials are applied inside the scenario and must be tackled to finish the tasks, I learned and practiced the script many times at home before performing.” A10 also noted, “Because the robot will ask me to do proper action or gesture and speak correctly, I seriously practiced and learned the script before performing in the classroom.”

Regarding “interesting,” all students reported that such a learning approach was fun and enjoyable. For example, A1 mentioned, “Learning with a robot inside the virtual world was more fun than the traditional teaching approach, thus making me interested in learning.” A6 answered, “The learning system is interesting, no pressure, and always makes the classroom atmosphere joyful, espe-cially due to the robot’s unexpected feedback.” Moreover, A9 stated, “I felt relaxed learning English by performing and watching the performances in the virtual world with many tasks from the robot given to students; thus, it made me almost forget that I was learning English.”

In terms of “fostering understanding,” eight students (A1, A2, A3, A5, A6, A7, A9, and A10) mentioned that learning with a robot in an interactive situated learning model enabled them to learn deeply and effectively. For instance, A3 revealed that “Performing with the robot in a journey deepens my understanding of the learning content. The learning system is helpful for me in solving real-world problems by using the English language.” Moreover, A2 and A7 further indicated that “The learning tasks, such as gesture, action, and keyword, that the performers should do are appropriate to the situation in the virtual scenario, thus making me understand the learning context.”

6. Discussion and conclusions

This study presented an interactive situated learning approach with a robot playing the role of character archetypes to perform, guide, and evaluate students during a situated learning activity. The proposed learning system allowed teachers to quickly transform learning content and contexts from textbooks into digital scenarios. Moreover, it allowed students to immerse in an authentic virtual world and interact with a robot, virtual objects, and virtual characters, with real-time evaluation and guidance. The combination of sensors and technologies from the robot and virtual world can generate immediate feedback and recognition capabilities, such as touch, gesture, and dialog, which enable evaluation of students’ knowledge application during performance, which is harder to achieve with human teachers alone. This study showed the potential of using a robot in a digital immersive setting. To evaluate the effectiveness of the proposed learning approach, a learning activity was conducted in a junior high school English as a second language course to compare the learning performances of students who learned with a robot in an interactive situated learning approach, situated learning approach, and conventional instruction approach.

The experimental results showed that the interactive situated learning group exhibited significantly higher learning achievement than the other two groups. Students learned better when the robot’s recognition capabilities (i.e., physical touch, embodied gestures, and dialogs) were added into the situational learning environment. The findings were consistent with previous research (Chang, Lee, Wang, & Chen, 2010; Kanda et al., 2004), which reported that adopting robots in interactive ways in the learning environment could enhance students’ learning performance. Furthermore, from the interviews of the students, it was inferred that the real-time situational feedback provided by the robot and the virtual world to evaluate and guide students could compel them to properly learn the script beforehand to perform well during the performance. Because they had to perform and complete tasks along their journey, they took responsibility for mastering the learning script. This process might increase the frequency of practice sessions. Students could learn more effectively if they were provided with adequate practice (Hwang & Wang, 2016). This could be a reason for why experimental

12

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Group A attained better learning achievement than the other groups. It was concluded that interactions and recognition provided by the proposed system could encourage students to study beforehand.

Interestingly, the students who learned in the conventional instruction group outperformed the situated learning group in terms of learning outcomes. This might be attributed to the students in the situated learning group not directly interacting with the robot but only interacting with virtual ones through a computer screen. Although the students in the conventional instruction group were not immersed in the virtual world, they directly interacted with the tangible robot. As reported in previous studies, physically embodied robots were more effective in producing learning gains (Belpaeme et al., 2018; Kose¨ et al., 2015; Leyzberg, Spaulding, Toneva, & Scassellati, 2012) than virtual robots. The physical embodiment of the robot played a significant role in enhancing students’ learning achievements; therefore, the physical presence of robots was an essential factor in interaction and learning in a classroom environment (Ahtinen & Kaipainen, 2020).

The learning motivation of students who learned with the robot in the interactive situated learning environment was significantly higher than that of the other two groups. This was because educational robot-based learning systems that combine tangible robots and attractive multimedia objects in the classroom could enhance learning motivation (Chin et al., 2014; Sugimoto, 2011). It was also found that both situated learning approaches, with and without interactive modes, resulted in significantly higher student learning motivation, as compared with the conventional instruction mode. Based on the study by Chen et al. (2018) and results of our interviews with the students, we believed that an authentic environment enhanced the experience and learning interest of the students, activating their learning motivation.

It was found that students who learned with the robot in an interactive situated learning environment outperformed the other two groups in terms of learning engagement. However, there were no significant differences in the learning engagement of the situated learning and conventional instruction groups. This could be attributed to the fact that students in the interactive situated learning group must overcome the challenges presented by the robot to obtain information and virtual objects or pass a gate into the next stage of the journey. This led to the performances being more engaging and fun, as stated by the students in the interviews. Several re-searchers have determined that tangible robots could help engage users (Jones & Castellano, 2018; Murphy et al., 2011); it was inferred that the support from the immediate evaluation and guidance present in the learning system led to a higher learning engagement (Yeh, Cheng, Chen, Liao, & Chan, 2019). Our findings corroborated that designing a robot in a situational learning environment with more interactions could help users engage better in contextual tasks.

Our study provides a novel area of insight into the use of robots in learning environments, by integrating digital technology with pedagogical approaches to generate an immersive virtual world and offer real-time evaluation and guidance. However, there were several limitations to this study. First, there was no baseline evaluation for learning motivation and engagement of experimental Groups A and B and the control group. Therefore, certain groups could initially be motivated and engaged in the activity. Second, we conducted our experiment on an English as a second language course for ninth-grade students. Consequently, we could not define the effect of our learning approach and system on students in other grades or for other subjects. Third, in our learning activity, developing a storyline approach based on school textbooks might be time-consuming in terms of the experimental time. Thus, researchers pre-pared the robot behavior and virtual world materials, such as interaction modes and virtual elements, to allow teachers to design the learning system quickly. We plan to allow users, both teachers and students, to design their own materials, including creating dialogs with themselves and having the robot act as professional speakers. Therefore, autonomy and ownership could hopefully emerge, supporting long-term use and engagement. Finally, we plan to propose audience participation during the performance by making the robot move into the audience area and interact with them. By doing so, the script would involve not only the student actors but also the audience. An appropriate dynamic script in an interactive situated learning environment is worth further investigation.

Credit author statement

Vando Gusti Al Hakim: Investigation, Writing – original draft, Writing – review & editing. Su-Hang Yang: Supervision, Resources, Visualization. Mahesh Liyanawatta: Software, Formal analysis. Jen-Hang Wang: Validation, Formal analysis. Gwo-Dong Chen: Conceptualization, Supervision, Methodology.

Funding

This work was supported by the Ministry of Science and Technology (MOST) of the Republic of China (R.O.C.) under contract numbers MOST: 109-2511-H-008-003-MY2, 109-2511-H-008-004-MY3, and 110-2811-H-008-505.

Acknowledgments

We would like to thank the teachers, especially Lung Yeh, and the Dalun Junior High School students for their involvement in this study.

References

Ahmad, M. I., Mubin, O., Shahid, S., & Orlando, J. (2019). Robot’s adaptive emotional feedback sustains children’s social engagement and promotes their vocabulary learning: A long-term child–robot interaction study. Adaptive Behavior, 27(4), 243–266. https://doi.org/10.1177/1059712319844182

13

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Ahtinen, A., & Kaipainen, K. (2020). Learning and teaching experiences with a persuasive social robot in primary school–findings and implications from a 4-month field study. In S. Gram-Hansen, T. Jonasen, & C. Midden (Eds.), Persuasive technology. Designing for future change. PERSUASIVE 2020. Lecture notes in computer science (pp. 73–84). Cham: Springer. https://doi.org/10.1007/978-3-030-45712-9_6.

Anderson, J. R., Reder, L. M., & Simon, H. A. (1996). Situated learning and education. Educational Researcher, 25(4), 5–11. https://doi.org/10.3102/ 0013189X025004005

Antle, A. N., & Wise, A. F. (2013). Getting down to details: Using theories of cognition and learning to inform tangible user interface design. Interacting with Computers, 25(1), 1–20. https://doi.org/10.1093/iwc/iws007

Babel, F., Kraus, J., Miller, L., Kraus, M., Wagner, N., Minker, W., et al. (2021). Small talk with a robot? The impact of dialog content, talk initiative, and gaze behavior of a social robot on trust, acceptance, and proximity. International Journal of Social Robotics. https://doi.org/10.1007/s12369-020-00730-0

Barab, S. A., Gresalfi, M., & Ingram-Goble, A. (2010). Transformational play: Using games to position person, content, and context. Educational Researcher, 39(7), 525–536. https://doi.org/10.3102/0013189X10386593

Barnes, J., FakhrHosseini, S. M., Vasey, E., Park, C. H., & Jeon, M. (2020). Child-robot theater: Engaging elementary students in informal STEAM education using robots. IEEE Pervasive Computing, 19(1), 22–31. https://doi.org/10.1109/MPRV.2019.2940181

Belpaeme, T., Kennedy, J., Ramachandran, A., Scassellati, B., & Tanaka, F. (2018). Social robots for education: A review. Science Robotics, 3(21). https://doi.org/ 10.1126/scirobotics.aat5954

Belpaeme, T., & Tanaka, F. (2021). Social robots as educators. OECD digital education outlook 2021: Pushing the Frontiers with AI, Blockchain and robots (pp. 143–157). Paris: OECD Publishing. https://doi.org/10.1787/1c3b1d56-en

Bosman, L., & Fernhaber, S. (2019). Applying authentic learning through cultivation of the entrepreneurial mindset in the engineering classroom. Education Sciences, 9

(1). https://doi.org/10.3390/educsci9010007. Article 7.

Bravo Sanchez,´ F.A´., Gonzalez´ Correal, A. M., & Guerrero, E. G. (2017). Interactive drama with robots for teaching non-technical subjects. Journal of Human-Robot Interaction, 6(2), 48–69.

Brown, J. S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18(1), 32–42. https://doi.org/10.3102/ 0013189X018001032

Chang, C. W., Lee, J. H., Chao, P. Y., Wang, C. Y., & Chen, G. D. (2010). Exploring the possibility of using humanoid robots as instructional tools for teaching a second language in primary school. Educational Technology & Society, 13(2), 13–24.

Chang, C. W., Lee, J. H., Wang, C. Y., & Chen, G. D. (2010). Improving the authentic learning experience by integrating robots into the mixed-reality environment. Computers & Education, 55(4), 1572–1578. https://doi.org/10.1016/j.compedu.2010.06.023

Chen, G. D., Fan, C. Y., Chang, C. K., Chang, Y. H., & Chen, Y. H. (2018). Promoting autonomy and ownership in students studying English using digital comic performance-based learning. Educational Technology Research & Development, 66(4), 955–978. https://doi.org/10.1007/s11423-018-9597-7

Cheng, Y. W., Wang, Y., & Chen, N. S. (2019). A framework for designing an immersive language learning environment integrated with educational robots and IoT-based toys. In M. Chang, et al. (Eds.), Foundations and trends in smart learning. Lecture notes in educational technology (pp. 1–4). Cham: Springer. https://doi.org/ 10.1007/978-981-13-6908-7_1.

Cheng, Y. W., Wang, Y., Yang, Y. F., Yang, Z. K., & Chen, N. S. (2020). Designing an authoring system of robots and IoT-based toys for EFL teaching and learning. Computer Assisted Language Learning, 34(1–2), 6–34. https://doi.org/10.1080/09588221.2020.1799823

Chen, H. R., & Lin, Y. S. (2016). An examination of digital game-based situated learning applied to Chinese language poetry education. Technology, Pedagogy and Education, 25(2), 171–186. https://doi.org/10.1080/1475939X.2015.1007077

Chen, H., Park, H. W., & Breazeal, C. (2020). Teaching and learning with children: Impact of reciprocal peer learning with a social robot on children’s learning and emotive engagement. Computers & Education, 150. https://doi.org/10.1016/j.compedu.2020.103836. Article 103836.

Chin, K. Y., Hong, Z. W., & Chen, Y. L. (2014). Impact of using an educational robot-based learning system on students’ motivation in elementary education. IEEE Transactions on Learning Technologies, 7(4), 333–345. https://doi.org/10.1109/TLT.2014.2346756

Chou, Y. S., Hou, H. T., Yu, M. C., Lee, H. J., Wu, H. S., Yang, Y. T., et al. (2012). March). Running Tommy©: Developing a digital adventure game based on situated learning to promote learners’ concepts of earthquake escape. In Proceedings of the 2012 IEEE fourth international conference on digital game and intelligent toy enhanced learning (DIGITEL2012) (pp. 156–158). https://doi.org/10.1109/DIGITEL.2012.44

Collentine, J., & Freed, B. F. (2004). Learning context and its effects on second language acquisition: Introduction. Studies in Second Language Acquisition, 26(2), 153–171. https://doi.org/10.1017/S0272263104262015

George, S., Michel, C., & Ollagnier-Beldame, M. (2016). Favouring reflexivity in technology-enhanced learning systems: Towards smart uses of traces. Interactive Learning Environments, 24(7), 1389–1407. https://doi.org/10.1080/10494820.2015.1016532

Gordon, G., Breazeal, C., & Engel, S. (2015). March). Can children catch curiosity from a social robot?. In Proceedings of the tenth annual ACM/IEEE international conference on human-robot interaction (pp. 91–98). New York: ACM. https://doi.org/10.1145/2696454.2696469.

Herrington, J., & Oliver, R. (1995). Critical characteristics of situated learning: Implications for the instructional design of multimedia. In J. Pearce, & A. Ellis (Eds.), Learning with technology (pp. 235–262). Parkville: University of Melbourne.

Hoffman, G. (2011). July). On stage: Robots as performers. In Proceedings of the RSS 2011 workshop on human-robot interaction: Perspectives and contributions to robotics from the human sciences (Los Angeles, CA).

Hwang, G. J., & Wang, S. Y. (2016). Single loop or double loop learning: English vocabulary learning performance and behavior of students in situated computer games with different guiding strategies. Computers & Education, 102, 188–201. https://doi.org/10.1016/j.compedu.2016.07.005

Hwang, G. J., Yang, L. H., & Wang, S. Y. (2013). A concept map-embedded educational computer game for improving students’ learning performance in natural science courses. Computers & Education, 69, 121–130. https://doi.org/10.1016/j.compedu.2013.07.008

Jamaludin, R., & Osman, S. Z. M. (2014). The use of a flipped classroom to enhance engagement and promote active learning. Journal of Education and Practice, 5(2), 124–131.

Jones, A., & Castellano, G. (2018). Adaptive robotic tutors that support self-regulated learning: A longer-term investigation with primary school children. International Journal of Social Robotics, 10(3), 357–370. https://doi.org/10.1007/s12369-017-0458-z

Kanda, T., Hirano, T., Eaton, D., & Ishiguro, H. (2004). Interactive robots as social partners and peer tutors for children: A field trial. Human-Computer Interaction, 19 (1–2), 61–84. https://www.tandfonline.com/doi/abs/10.1080/07370024.2004.9667340.

Kennedy, J., Baxter, P., Senft, E., & Belpaeme, T. (2016). March). Social robot tutoring for child second language learning. In 2016 11th ACM/IEEE international conference on human-robot interaction (HRI) (pp. 231–238). IEEE. https://doi.org/10.1109/HRI.2016.7451757.

Konijn, E. A., & Hoorn, J. F. (2020). Robot tutor and pupils’ educational ability: Teaching the times tables. Computers & Education, 157. https://doi.org/10.1016/j. compedu.2020.103970. Article 103970.

Kose,¨ H., Uluer, P., Akalın, N., Yorgancı, R., Ozkul,¨ A., & Ince, G. (2015). The effect of embodiment in sign language tutoring with assistive humanoid robots. International Journal of Social Robotics, 7(4), 537–548. https://doi.org/10.1007/s12369-015-0311-1

Kosmas, P., Ioannou, A., & Zaphiris, P. (2019). Implementing embodied learning in the classroom: Effects on children’s memory and language skills. Educational Media International, 56(1), 59–74. https://doi.org/10.1080/09523987.2018.1547948

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge: Cambridge University Press.

Legault, L., Green-Demers, I., & Pelletier, L. (2006). Why do high school students lack motivation in the classroom? Toward an understanding of academic amotivation and the role of social support. Journal of Educational Psychology, 98(3), 567–582. https://doi.org/10.1037/0022-0663.98.3.567

Lemaignan, S., Gharbi, M., Mainprice, J., Herrb, M., & Alami, R. (2012). March). Roboscopie: A theatre performance for a human and a robot [video abstract]. In 2012 7th ACM/IEEE international conference on human-robot interaction (HRI) (pp. 427–428). New York: ACM. https://doi.org/10.1145/2157689.2157831.

Leyzberg, D., Spaulding, S., Toneva, M., & Scassellati, B. (2012). The physical presence of a robot tutor increases cognitive learning gains. Proceedings of the Annual Meeting of the Cognitive Science Society, 34(34), 1882–1887. https://escholarship.org/uc/item/7ck0p200.

14

V.G. Al Hakim et al.Computers & Education 182 (2022) 104483

Liyanawatta, M., Yang, S. H., Liu, Y. T., Zhuang, Y., & Chen, G. D. (2021). Audience participation digital drama-based learning activities for situational learning in the classroom. British Journal of Educational Technology. https://doi.org/10.1111/bjet.13160

McCombs, B. L. (2000). Reducing the achievement gap. Society, 37(5), 29–36. https://doi.org/10.1007/s12115-000-1034-x

Murphy, P. K., & Alexander, P. A. (2000). A motivated exploration of motivation terminology. Contemporary Educational Psychology, 25(1), 3–53. https://doi.org/ 10.1006/ceps.1999.1019

Murphy, R., Shell, D., Guerin, A., Duncan, B., Fine, B., Pratt, K., et al. (2011). A Midsummer Night’s Dream (with flying robots). Autonomous Robots, 30(2), 143–156. https://doi.org/10.1007/s10514-010-9210-3

Okita, S. Y., Ng-Thow-Hing, V., & Sarvadevabhatla, R. (2009). September). Learning together: ASIMO developing an interactive learning partnership with children. In

Proceedings of the 18th IEEE international symposium on robot and human interactive communication (pp. 1125–1130). https://doi.org/10.1109/ ROMAN.2009.5326135

Rajendran, R., Iyer, S., & Murthy, S. (2018). Personalized affective feedback to address students’ frustration in ITS. IEEE Transactions on Learning Technologies, 12(1), 87–97. https://doi.org/10.1109/TLT.2018.2807447

Saerbeck, M., Schut, T., Bartneck, C., & Janse, M. D. (2010). April). Expressive robots in education: Varying the degree of social supportive behavior of a robotic tutor. In Proceedings of the SIGCHI conference on human factors in computing systems (pp. 1613–1622). New York: ACM. https://doi.org/10.1145/1753326.1753567.

Sugimoto, M. (2011). A mobile mixed-reality environment for children’s storytelling using a handheld projector and a robot. IEEE Transactions on Learning Technologies, 4(3), 249–260. https://doi.org/10.1109/TLT.2011.13

Sung, H. Y., Hwang, G. J., & Yen, Y. F. (2015). Development of a contextual decision-making game for improving students’ learning performance in a health education course. Computers & Education, 82, 179–190. https://doi.org/10.1016/j.compedu.2014.11.012

Viljaranta, J., Lerkkanen, M. K., Poikkeus, A. M., Aunola, K., & Nurmi, J. E. (2009). Cross-lagged relations between task motivation and performance in arithmetic and literacy in kindergarten. Learning and Instruction, 19(4), 335–344. https://doi.org/10.1016/j.learninstruc.2008.06.011

Vogler, C. (2007). The writer’s journey. Studio city: Michael wiese productions.

Yeh, C. Y., Cheng, H. N., Chen, Z. H., Liao, C. C., & Chan, T. W. (2019). Enhancing achievement and interest in mathematics learning through Math-Island. Research and Practice in Technology Enhanced Learning, 14(1). https://doi.org/10.1186/s41039-019-0100-9. Article 5.

Yeoman, P., & Wilson, S. (2019). Designing for situated learning: Understanding the relations between material properties, designed form and emergent learning activity. British Journal of Educational Technology, 50(5), 2090–2108. https://doi.org/10.1111/bjet.12856

Zhang, Z. (2012). Microsoft Kinect sensor and its effect. IEEE multimedia, 19(2), 4–10. https://doi.org/10.1109/MMUL.2012.24

15