A Study on Elementary English Teachers’ Responses to Classes Using AI-Powered Application: Focusing on Odinga English
Article information
Abstract
The purpose of this study is to examine the responses of English teachers to utilizing an AI-based interactive application, Odinga English. To determine their views, a case study was conducted with four English teachers. The findings are as follows. First, the contents of the application consisted of fun elements with graphics which had a positive effect on increasing learners’ interest and motivation. In addition, learners improved their self-confidence by solving problems on their own. Second, the design and interface of the application were appealing to young learners, an important point as both accessibility and convenience play an important role when lower-grade students are learning with a new tool. Third, the adaptation of the educational application was appropriate for elementary school students who are familiar with using technology to learn. Lastly, each learner received individual feedback on their speaking and pronunciation. Based on the results, it can be concluded that the use of the AI-based application was positive for elementary school English learners. Above all, it was effective in increasing learning interest and confidence, which are important for elementary school learners. It was also positive in improving learners’ speaking and pronunciation skills through customized feedback.
I. INTRODUCTION
Entering the era of the next industrial revolution, various artificial intelligence chatbot programs applied with voice conversation processing technology have been developed and used for various purposes. However, the development of a program that combines voice conversation processing technology and artificial intelligence technology that can be used in foreign language (English) education is still very lacking in quality and quantity. Some programs have provided pronunciation evaluation services using speech recognition, but the development of programs that provide effective feedback or include technology related to conversation processing is still insufficient. Despite the development of voice conversation processing technology, there are many difficulties in providing foreign language learning services due to the absence of teaching and learning methods suitable for the educational area. Therefore, it may be urgent to develop programs equipped with technologies that can be effectively used in the language class and to develop appropriate teaching and learning methods at the same time. In addition, if the effectiveness of the developed teaching and learning method is proven through verification, it is expected that various voice conversation processing technology-based language teaching and learning methods can be provided to the educational field. In order to reflect this trend, this study focused on effective ways to use voice conversation processing technology in language teaching and learning environments.
In the case of Korea, looking at the 2015 revised curriculum, the total number of regular English classes for students during the 10 years of elementary, middle, and high school is approximately 1,100 hours. This is significantly less than the average number of hours required to reach the level 3 ACTFL1 standard, which requires more than 3,000 hours (Hardley, 2001). The lack of class hours not only means a lack of opportunities to develop listening and reading skills which are receptive, but also a lack of opportunities to improve speaking and writing skills corresponding to productive ones. In this regard, the number of classes in the English curriculum in Korea for boosting communicative competence needs to be changed. In addition, for the balanced development of 4 skills of language including receptive and productive ones, an environment in which learners can actively access English is needed.
In the English curriculum in Korea, elementary English education generally focuses on oral language than written language, and the overall learning contents are organized in connection with oral language. From the 2011 revised national curriculum, the standards previously presented by grade group were divided into 3-4th and 5-6th grades for elementary schools, and 1-3rd graders for middle schools into one grade group to set content elements and achievement standards for each area. Achievement standards for each grade, and learning content were clearly and specifically presented, and major learning activities were provided as examples. In addition, achievement standards for each level within the grade were presented in consideration of the connection between school levels. The biggest feature of the 2011 revised English curriculum was that the imbalance in language understanding (listening and reading) and expression (speaking and writing), which had been the most problematic in Korean English education, has been balanced to some extent through elementary English education standards. In language development and acquisition, it is very important that learners are provided with balanced input and output opportunities, allowing them to develop the 4 skills of language integrally and naturally. In this regard, in the 2011 revised English curriculum, the balanced presentation of listening, speaking and reading, and writing achievement standards of elementary English achievement standards was particularly meaningful.
However, in the 2015 revised English curriculum for the elementary English standards, the balance of input and output in the areas of listening and speaking, reading and writing tended to return to before the 2011 revised English curriculum. As the level of achievement in the listening and reading areas of input-oriented was strengthened as in the previous curriculum, the balance between the level of achievement of speaking and writing, which were relatively expressive functions, was being broken. The level of listening and reading in the revised elementary English curriculum for 3-4th graders has been partially lowered to the level of achievement for 5-6th graders, while the level of speaking and writing remained the same for 3-4th graders in the 2011 elementary English curriculum. As such, the imbalance in the achievement standards of language 4 skills with strengthened input functions and weakened output functions became a very serious problem, and this problem was also seen in the 2015 revised elementary English and 5-6th grade standards. This imbalance in the standards of receptive and productive skills continued in the middle and high school English curricula. Although the balanced development of receptive and productive skills is very important in the process of language development, English curriculum in Korea did not reflect this in depth. And this means more research on teaching and learning methods that can develop productive skills should be needed. Therefore, the purpose of this study is to verify an effective teaching-learning method in elementary English education to integrate 4 main skills and specifically to develop language learners’ output. To this end, this study focused an English interactive learning application equipped with voice conversation processing technology, which was designed to escalate the learners’ oral skills, and the responses of English teachers through their empirical reports were examined to figure out any educational effects, merits or demerits for language teaching and learning.
II. LITERATURE REVIEW
1. Overview of Voice Conversation Processing Technology
Speech recognition refers to dividing pronounced speech into basic units, identifying them, and finally deriving meaning through linguistic analysis and outputting them back into speech (Choi et al., 1991). And voice conversation processing technology basically enables voice interaction with learners based on computers. With the advent of the 4th Industrial Revolution, innovative advances in information communication technology (ICT) have led to rapid development in speech language recognition, and foreign language technologies such as automatic speech recognition (ASR), computer speech recognition (CSR), or text to speech conversion (TTS). The core of this technology is that the input of human voice interface is extracted and output as conversation or text data, which operates based on voice recognition and natural language understanding. Acoustic models are responsible for converting human speech sounds into text by converting them into a series of digital signals, and language models are models of language rules, helping to properly recognize texts obtained through acoustic models. Figure 1 is a process of voice conversation processing technology.
Voice Conversation Module
According to Kwon et al. (2015), spontaneous speech dialogue processing (SSDP) technology is to continue conversations through flexible responses depending on situations and contexts, away from fixed conversations in simple patterns between humans and computers. In other words, it can be said that it is a technology that enables a computer to grasp the intention of human natural speech and to have a natural conversation according to the context. Representative examples of the application of voice language recognition technology include Google Assistant, Apple’s Siri, and navigation installed in smart cars such as self-driving cars. Speech conversation processing is basically approached from a phonetic and linguistic perspective, and both Google Assistant and Apple’s Siri work both based on human speech (voice) and language data such as context, grammar, and vocabulary. The process of processing voice language outputs text information from voice signals and then interprets the speaker’s intention through the process of understanding natural language, which is called decoding. Subsequently, the input information is understood through decoding, and then the voice is output in consideration of acoustic features and grammar systems. Speech synthesis technology is applied to text to speech, and speech synthesis is the process of converting a series of phonetic and prosodic symbols into human speech using synthesized speech (Kang, 2004). Text-to-speech conversion includes a text module that acoustically analyzes input text and a voice module that implements actual speech sounds through deep learning and machine learning. Just as voice recognition utilizes voice recognition technology to provide related information, interactive voice recognition technology with an interactive interface added to the voice recognition base has recently been introduced (Hong et al., 2020).
2. Research About Voice Conversation Processing Technology
Current research studies related to voice conversation processing technology mainly focused on cases using chatbots, and many results have been introduced as related topics within the past three years. Go and Lee (2020) introduced a chatbot service using a messenger interface for Chinese education in Korea, prepared a chatbot design plan for Chinese learning, and proposed a teaching-learning model applicable to Chinese language education field. Kim et al. (2020) utilized chatbots using the KakaoTalk open builder, and through this, positively recognized the possibility of using chatbots systematically learned through deep learning. Examples of research related to English speaking, listening, and evaluation using chatbots for elementary English learners can be examined by Min (2019), Shin (2019), Shim et al. (2020), Yang et al. (2019), Lee (2019), Chu and Min (2019). First of all, Min (2019) reported that the overall conversation success rate exceeded 80% through the small talk sharing task led by learners as a result of elementary English classes using chatbots for elementary school students. Shin (2019) also cited the advantage of chatbot in English education in that it provides opportunities for learners to receive grammatical feedback and conduct English learning using natural colloquial expressions through English classes using Mitsuku (currently called Kuki AI)2. In addition, the fact that it was more psychologically comfortable than conversation with humans has great implications for learning English using chatbots. In order to use AI chatbots for elementary English evaluation, Shim et al. (2020) devised a process-oriented evaluation model and developed a chatbot suitable in English class. As a result, it was reported that it was possible to observe the improvement of learners’ English skills and the integration of learning and evaluation. Yang et al. (2019) conducted an experimental study on whether elementary English learners achieved conversation tasks through a chatbot called Ellie produced by Dialogflow. As a result of the study, it was confirmed that the use of customized chatbots by learner level was effective in improving conversation ability. Lee (2019) conducted an experiment on the interest, attitude, and confidence of elementary English learners using chatbots, and reported that gamification-based AI chatbot classes had a positive effect on the affective factors of elementary English learners. Chu and Min (2019) explored the possibility of using a user-centered chatbot that continuously induces English speaking.
Based on the previous studies, use of AI chatbot in English education showed potentials which would be practically possible by providing English language learners with fair learning opportunities at various levels, that is, personalized education. Choi (2020) said that the AI chatbot’s specific function (history) serves as an advantage of recording students’ speaking processes in course-based speaking evaluation, and this function could confirm not only the results of the evaluation, but also provide very useful information for students’ English education in the future. There were also some results that chatbots had a positive effect on not only English learners’ cognitive achievement but also on improving their immersion, motivation, and confidence in classes. Kim and Kim (2020) said that the use of chatbots in language education is effective in improving learners’ language performance and motivation. A study by Sung (2020), Sung and Kang (2020) made a similar argument, confirming that chatbots are an excellent medium for improving English learners’ cognitive and affective abilities, as well as improving English teachers’ expertise and capacity. However, despite the advantages and educational possibilities of using chatbots as above, there are also research results that suggest practical problems and points to be improved. Studies by Son et al. (2013), Jung (2019), Choi and Nam (2019), and Ki (2019) argued that despite the potential and educational effects of chatbots, practical help such as infrastructure construction, teacher education, and model development of English classes through conversation processing technology should be accompanied.
III. METHOD
1. Participants
This study was conducted in after-school English classes at 4 different public elementary schools in 2021. In order to explore the validity and effectiveness of the teaching and learning tool, participating schools were selected by the judgment sampling, and the responses of teachers were investigated to grasp their views and insights on using AIbased English teaching and learning tool. To this end, Odinga English, an artificial intelligence-based English learning tool, was introduced and used, and detailed information on the number of participants is shown in Table 1.
Participants
2. Learning Tool
The main teaching and learning tool used for this study is Odinga English3 application, an artificial intelligencebased English learning tool. It was of similarity to an AI-powered English speaking practice application, EBS AIPeng Talk4 in public education, and as a voice recognition and AI-based English conversation application, it had features to allow young English learners to practice and speak English effectively by themselves, reflecting high motivation, language self-esteem, and low anxiety as the important factors in second language learning. The main features of the application are shown in Figure 2 below.
Screenshots of Odinga English Application
Odinga English provided learners with patterns practice, webtoon role-play, vocabulary learning, YBM webtoon from simple expressions to various complex ones and conversation practice. It was grounded on the gamification which allows young children to learn English like a game and keeps them motivated, engaged and interested. After-school classes were held for 15 weeks for 30 minutes, and teachers used the Odinga application as a main tool to conduct speaking-listening activities with learners.
3. Procedures
As mentioned above, 4 number of elementary English teachers in 4 different public schools in Incheon and Jeolla province taught after-school English classes using the AI-powered application, Odinga English. Teachers were of average 7-year-teaching experience and were involved in the lesson using Odinga for 15 weeks. One session consisted of 30 minutes, and about half of them was targeted learners to practice speaking and listening in general.
Procedures
4. Data Collection and Analysis
In order to collect data for this study, empirical research activity reports of the focal English teachers were administered and gathered and then analyzed to understand their responses and perspectives in using AI-based application (see Appendix). An empirical report was composed of open-ended question type and was provided when the experiment was over. In particular, to explore the effectiveness of foreign language teaching and learning method using voice conversation processing technology, affective domains (motivation, interest, confidence, attitude) were analyzed on changes in speaking and listening skills. It also identified the efficiency for device utilization and user convenience, and finally summarized the expected effects and directions for improvement. Table 3 below summarizes the empirical research activity report organization.
Organization of Empirical Research Activity Report
IV. RESULTS
1. Changes of Affective Domains
4 English teachers’ empirical reports were analyzed to examine how speaking-listening activities using Odinga English application affected any changes of the affective domain of elementary English learners. First of all, according to Teacher A, the contents of the Odinga application was fun and appealing, which had a positive effect on learning motivation. In addition, learners who had difficulty speaking were able to solve problems on their own and gain confidence through activities in which learners became teacher’s role. In terms of speaking attitude, it was observed that most of the students tried to pronounce confidently as they became familiar with speaking activities over time and their fear decreased. In this regard, Teacher B’s opinion was as follows.
Children who were shy and hated to speak English actively tried to speak English and naturally learned English by listening and following the pronunciation repeatedly. (Teacher B)
Class Activities With Odinga English
Besides, based on the gamification effect that combined fun and game elements with learning, Teacher B judged that speaking-listening activities through the application had a positive effect on the affective domain of young learners.
2. Changes of Utilization
Following the changes in the affective domain of learners, teachers’ opinions on the utilization of application were dealt with. First of all, Teacher C said that as a result of examining the design and interface of the product, it partially matched the level of learners and at the same time had a good effect on learning motivation. Regarding learning operability, there was no difficulty in accessing Galaxy Tab, iPad, or general smartphones, but it was expected that access might be difficult or challenging depending on the network environment. Teacher C stressed that both accessibility and convenience would play an important role, especially when lower grade students learn with new devices.
However, it was pointed out that a system enabling to check and monitor the progress of each learner’s learning would be added through the teacher’s account. In particular, as there was no function to monitor information related to improving learners’ accurate pronunciation, it also would need to supplement.
3. Changes of Efficiency
As for the efficiency of using Odinga English, Teacher C cited the effectiveness of device use as an advantage in that it can easily access the application at the elementary school level and proceed with learning on its own. In terms of education connectivity, from easy to difficult stages, it was linked to the contents of the current elementary school curriculum. However, there was a slight lack in learning expressions actually used in schools because the contents of textbooks of all publishers were not reflected. In providing feedback, the Odinga application provided customized feedback based on the learner level, and properly provided feedback by catching individual sounds even when many students were talking at the same time.
4. Expectations and Directions
After about 15 weeks of English speaking-listening practice through an AI-based Odinga, the participating teachers generally showed a few important educational suggestions. The analysis of the teachers’ perception is as follows. First, the teachers recognized Odinga English equipped with thorough voice conversation processing technology as an auxiliary vehicle to achieve learning goals. If the application was designed to suit the level of elementary learners and contained expressions of words and sentences presented in elementary English textbooks, it could have been actively integrated into the class and operated. Second, it is about technical limitations. The teachers, who observed the participants’ learning during the class, assumed that when the technical completeness of an interactive English learning application equipped with voice conversation processing technology was guaranteed, it could be used as an efficient tool for the class. Considering that learners were very immersed in learning the target language through various activities provided by the application, learners quickly became less immersed and felt negative emotions when the application showed many technical and systematic errors. Therefore, in order for learners to maximize the learning effectiveness using the application, technical completeness must be guaranteed before the aesthetic and content suitability of the application. In addition, when an error is found in the program, a bypass path to cope with it needs to be prepared in terms of teaching and learning. In fact, teachers agreed that learners did not rely on the use of the application itself, but used the application in the process of linking the teachers’ input activity and subsequent output activity. This was also a measure considering the imperfections of the application, but it is worth noting that it was a teaching and learning measure to maximize the effectiveness of language learning through input, interaction, and output activities that learners would feel in the language learning process. Third, there was an issue of feedback. It has been mentioned earlier that in order for the feedback provided by the application to become meaningful information for learners, it would be necessary to supplement it in the process of designing the application. At the same time, in offline classes, teachers should interpret the information provided by the application to individual learners and explain it in detail so that learners can use it, and at the same time, praise and encouragement should be given to actively participate in speaking activities. Finally, it was found that teachers were concerned about the overall activities in the teaching and learning process that encompassed application activities. Since activities were mainly characterized by individual learners, it is argued that there is a need for additional activities to be carried out that can be applied and linked through interaction activities between learners after sufficient speaking practice activities individually. However, in spite of several challenges, the teachers’ responses also indicated adapting a new tool has the potential that English learners can accelerate their communication skills, enhance their active participation in English learning, and even promote self-directed learning.
V. CONCLUSIONS AND IMPLICATIONS
English subjects have traditionally been guided around instructors using textbooks, but it may be urgent to change a new teaching-learning paradigm to achieve the goals of foreign language (English) subjects set by the national curriculum. Future learners, called digital natives, are sensitive to education using digital devices. In addition, in an EFL environment where English is learned as a foreign language like Korea, exposure and frequency of use of target words are absolutely important. In other words, as practical language use is limited except for the classroom environment, the incorporation of voice conversation recognition technology using artificial intelligence can go beyond the constraints of the EFL environment and become an excellent alternative to future learners. It will be a good opportunity to improve communication skills through creative and interactive practice, away from simple repetition and imitation training. To this end, continuous development and upgrade of new tools that positively affect not only the learner's cognitive domain but also the affective domain of English learning are essential.
In view of these points, this study aims to verify an effective teaching-learning method in elementary English education using voice conversation processing technology. To answer this question, an English interactive learning program equipped with voice conversation processing technology was applied and examined the responses of English teachers. The findings are as follows. First, the contents of the Odinga application consisted of fun elements with graphics which had a positive effect on escalating learners’ interest and motivation. In addition, learners improved their self-confidence by solving problems on their own. Second, the design and interface of the product were appealing to young learners, and both accessibility and convenience played an important role, especially when lowergrade students learned with a new resource. Third, the adaptation of the educational application was proper for the elementary school students who have been aware of using such applications during the pandemic. Lastly, properly receiving feedback on learners’ speaking and pronunciation based on each individual was effective.
To sum up, it can be concluded that the use of AI-powered application played a positive function for elementary English learners. Above all, it was effective in increasing learning interest and confidence, which are very important factors for elementary school learners, and it was also positive in improving learners’ speaking and pronunciation skills through customized feedback. Finally, there are some limitations to this study. First of all, there is difficulty in generalizing the contents of their empirical reports because there were too few participants. At the same time, it may be difficult to generalize the results because only schools in certain regions were selected and experimental classes were conducted. Nevertheless, as there have been few cases of investigating teachers’ perceptions and responses in English classes using artificial intelligence-based tools. In this respect, this study may be of importance.
Notes
As of the 2021 academic year, users except for classroom teachers in a public school did not gain permission to use it.