Computers and Education Artificial Intelligence

Computers and Education: Artificial Intelligence 4 (2023) 100134
Contents lists available at ScienceDirect
Computers and Education: Artificial Intelligence

journal homepage: www.sciencedirect.com/journal/computers-and-education-artificial-intelligence
A systematic review on artificial intelligence dialogue systems for

enhancing English as foreign language students’ interactional competence
in the university
Chunpeng Zhai a, Santoso Wibowo b
a
School of Engineering and Technology, Central Queensland University, Brisbane, Australia
b
School of Engineering and Technology, CQUniversity Australia, 120 Spencer Street, Melbourne, VIC 3000, Australia
A R T I C L E I N F O A B S T R A C T
Keywords: Previous studies demonstrate that the use of artificial intelligence (AI) dialogue systems for English as a Foreign
Artificial intelligence Language (EFL) education has effectively improved university students’ reading, writing, and listening abilities.
Dialogue system However, there are limited systematic reviews focused on the evidence-based interactional competence of EFL
Foreign language learning
university students. This study aims to examine the use of AI dialogue systems to enhance EFL university stu
Interactional competence
University education
dents’ interactional competence. Through the PRISMA process, this study identified 28 articles published be
tween January 2013 and August 2022 in journals and conferences from the most popular databases, including
Google Scholar, ProQuest, IEEE, ScienceDirect, and Web of Science. The systematic review identified six di
mensions and 25 sub-dimensions that influence the application of AI dialogue systems for EFL learning. The six
dimensions include technological integration, task designs, students’ engagement, learning objectives, techno
logical limitations, and the novelty effect. Gaps are identified that (1) components of debate and problem-solving
skills in EF acquisition in university education seemed to be overlooked in the AI dialogue system design, and (2)
the importance of embedding culture, humor and empathy functions were not taken into consideration in the AI
dialogue system. This study finds that the development and implementation of an AI dialogue system in EFL is
still in its infancy stage. Future research should emphasize meaning-based communication, intelligibility in
language competency, debate, and problem-solving skills in university education.
1. Introduction between 1964 and 1966 as the first natural language processing com
puter software. Eliza was built to highlight the superficiality of human-
An artificial intelligence (AI) dialogue system is a software applica machine interaction via the use of pattern matching to imitate conver
tion that simulates natural human dialogue through the use of text or sation (Weizenbaum, 1976). The revolutionary ELIZA paved the way for
text-to-speech functions. This is accomplished with the assistance of other projects on chatbots, which are also known as text-based dialogue
natural language processing (NLP), which can be achieved through systems (Mastura, 2021). The introduction of AI Markup Language
machine learning or deep learning (Zhai & Wibowo, 2022). Over the last (AIML) has facilitated the design of dialogue systems for EFL learning.
decade, AI dialogue systems have been gathering momentum as studies Some practitioners of computer-assisted language learning (CALL) can
show that these dialogue systems are easy to use, unbiased and in return use established dialogue systems by appraising their language-learning
boost users’ confidence in a more friendly setting (Alsadoon, 2021; El potential based on the availability of an open-source framework (Mas
Shazly, 2021; Hsu, Chen, & Todd, 2021). Nowadays, AI dialogue sys tura, 2021). Due to the wide developer community, other CALL practi
tems are used in a variety of domains, including health, marketing, tioners are able to construct additional systems for language students,
business, retail, entertainment and foreign language learning (Hsu et al., such as computer simulators in educational communication (CSIEC)
2021). (Mastura, 2021).
The development of novel language-learning applications can be In the field of foreign language acquisition, several applications
traced back to the first dialogue system, ELIZA. ELIZA was developed by using AI dialogue systems have been developed to create interactive
Joseph Weizenbaum at the MIT Artificial Intelligence Laboratory tasks for enhancing various aspects of a language learner’s interactional
E-mail address: [email protected] (C. Zhai).
https://doi.org/10.1016/j.caeai.2023.100134
Received 12 September 2022; Received in revised form 18 March 2023; Accepted 18 March 2023
Available online 24 March 2023
2666-920X/© 2023 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
C. Zhai and S. Wibowo Computers and Education: Artificial Intelligence 4 (2023) 100134
competence (Mastura, 2021; Timpe-Laughlin, Sydorenko, & Daurio, enhance interactional competence among Thai students of EFL in
2020; Young, 2011). Interactional competence is the capacity to use one-on-one tutorials. The findings suggest that 76 per cent of them
available language resources to deploy interactional resources as improved their speaking skills, and 100 per cent of them identified
required by the speaker/hearer to convey their communication intents convergence as the most prevalent function of listenership in order to
in real situational circumstances (Alsadoon, 2021). Interactional demonstrate that they understood the material and to begin to converse.
competence is the core of the communication abilities that EFL students However, existing studies found that university students in many EFL
are expected to acquire (Young, 2011). In his theory of interactional settings do not have many opportunities available to enhance their
competence, Young (2011) conceptualizes interactional communication interactional competence outside of the classroom or even inside the
as including but going beyond the components of communicative classroom, as the spoken interactional practice is frequently limited to
competence. These components include: (a) rhetorical script; (b) the interactions between the teacher and classmates (Wahyuningsih &
register of the practice; (c) the turn-taking strategies; (d) topic man Afandi, 2020; Xie, 2017; Zrekat & Al-Sohbani, 2022).
agement; (e) the participation framework; and (f) means for signaling An AI dialogue system for EFL is anticipated to play a significant role
boundaries and transitions (Skogmyr & Balaman, 2018). In more eth in influencing EFL learning at the university level in the near future
nomethodologically oriented research within the field of interactional (Roever & Kasper, 2018). Moreover, as the AI EFL dialogue system is
competence, researchers have focused on the observable, micro-level commonly viewed as a tool that can leverage technical capacities to
interactional aspects of speakers’ competences (Girgin & Brandt, resolve difficult and complex foreign language learning problems, a
2020; Hall, Hellermann, & Doehler, 2011; Skogmyr Marian & Balaman, growing body of studies is gaining traction on how AI dialogue systems
2018). for EFL can be used to support university students’ interactional
competence (Rosé & Ferschke, 2016). This development will affect how
1.1. Interactional competence EFL students frame and examine issues, analyze multiple perspectives,
and apply various resources via collaboration and interactions. Such
Interactional competence entails understanding communication rules learning environments require that EFL university students cultivate
as well as utilization of both linguistic and interactional resources, such as enhanced agency and cognitive, social-emotional, and behavioral skills
turn-taking and task-related management, within a particular context for functioning in knowledge societies.
(Galaczi & Taylor, 2018). In EFL spoken interactions, two interlocutors The main goal of EFL learning for students is arguably to acquire
participate and co-construct in vigorous and reciprocal conversations fundamental language skills to enhance the performance of communi
(Galaczi & Taylor, 2018). Interactional competence reflects a type of cative competence (Young, 2011). The term competence was first
interaction, such as collaborative and asymmetric, created by in introduced by Chomsky (1965) in the field of linguistics in order to
terlocutors in conversations by drawing from two perspectives: differentiate between knowledge of language in the abstract (compe
sociolinguistic-interactional and psycholinguistic-individualist perspec tence) and the way in which knowledge is realized in the production and
tives (Roever & Kasper, 2018). interpretation of actual utterances. However, Hymes (1972) challenged
A psycholinguistic-individualist perspective focuses on an emphasis Chomsky’s conception of competence, which he defined as knowledge of
on the individual speaker and their communication skills, and it eval language apart from its use.
uates combinations of criteria such as fluency and pronunciation, lexical The concept put out by Hymes (1972) served as the foundation for an
resources and grammatical variety, and topic management. However, applied linguistic theory of communicative competence that was
the psycholinguistic-individualist perspective does not require that developed by Canale and Swain (1980). This theory tied language ac
language usage be integrated in social interaction or situational settings tions that took place in social circumstances to underlying knowledge. In
(Roever & Kasper, 2018). This is a critically problematic for the applied linguistics, language testing, and language education, commu
real-world application of a language, since interlocutors do engage in a nicative competence was considered to be a feature of a particular
range of contexts and with a variety of interlocutors, and not merely to person. This complex construct consisted of various component ele
generate context-free fluent, correct, and complicated language. ments that separated one individual from others and made up commu
A sociolinguistic-interactional perspective places an emphasis on the nicative competence. Young (2011) proposed the interactional
aims of speakers’ linguistic choices as well as the impacts those choices competence theory as an alternative theoretical framework to commu
have on their interlocutors (Roever & Kasper, 2018). It evaluates the nicative competence.
appropriateness, conventionality, and efficacy of their utterances. It is The last decade has seen an unparalleled and exponential rise in the
reflective of sociolinguistic conceptions of language competency, with a capabilities of technology and AI, which is posing a challenge to tradi
concentration on the contextual and social dimensions of language tional university education (Moscardini, Strachan, & Vlasova, 2022).
usage. The sociolinguistic-interactional perspective is deemed to The use of an AI dialogue system for language learning is relatively new
enhance interactional competence in real-world language usage that as students need to obtain trust and positive attitudes towards an EFL
involves the ability to deploy language in interactive circumstances for dialogue system (Bashori, van Hout, Strik, & Cucchiarini, 2020).
social purposes and academic language (Roever & Kasper, 2018). Engaging in interactional practices in the target language with dialogue
Researchers have applied the interactional competence theory to systems can also reduce anxiety and increase the desire to participate in
various language learning contexts, such as administering a paired oral speaking practice (Bashori et al., 2020). There are prior studies on the
discussion task within the context of language testing (Divekar et al., use of dialogue systems to replicate a virtual target language-speaking
2021, pp. 1–29; Huth & Betz, 2019; Kampittayakul, 2019). Through environment. This was accomplished via two different kinds of
implanting these activities, the researchers were able to examine the test learning activities: (a) a role-play activity and (b) a learning scenario
taker’s knowledge and abilities situated in the sequentially of interac representation (Huang, Hew, & Fryer, 2022; Wollny et al., 2021).
tion and how learners can negotiate. The studies found that learning However, there are several major issues relating to EFL learning for
goals in the realm of interactional competence can be effectively taught university students such as:
and learnt in EFL class. In one study, Divekar et al. (2021, pp. 1–29)
leveraged the interactional competence theory to engage students of (a) Lack of focus on improving EFL students’ long-term learning
foreign language learning in non-dyadic multimodal conversations objectives from spontaneous initiation of engagement with the
within the cognitive immersive language learning environment. The dialogue systems,
findings demonstrate a statistically significant and sustained increase in (b) Overlooked factors enhancing EFL students’ wellbeing in the
vocabulary, understanding, and conversational abilities. Meanwhile, design of dialogue systems for EFL during academic life at
Kampittayakul (2019) investigated the use of translanguaging to university,
2
(c) Lack of cross-cultural humor consideration in the dialogue sys questions around them, researchers can better understand the effec
tems for EFL, tiveness of AI dialogue systems for EFL and contribute to the ongoing
(d) Lack of cross-cultural awareness function in designing dialogue development of these systems in university settings. To address the
systems for EFL, and existing gap and contribute to the comprehension of how AI dialogue
(e) Absence of cross-cultural empathy practice in designing dialogue systems have been utilized to enhance the interactional competence of
systems for EFL. EFL students, this systematic review analyzed studies that investigated
interactional competence through the use of AI dialogue systems. This
How might universities, lecturers, researchers, training developers, systematic review is guided by the research questions listed below:
and EFL dialogue system developers contribute to the advancement of
EFL university students’ interactional competence? Answering this 1. What are the key evaluation dimensions of AI dialogue systems for
question requires a more in-depth examination of how research studies enhancing student’s EFL interactional competence?
with existing AI dialogue systems for EFL which are enhancing the 2. What gaps arise from the results, and what additional study is
interactional competence of EFL students and requires to conduct required for achieving effective AI dialogue systems?
research studies with existing AI techniques. One important aspect to
consider in evaluating AI dialogue systems for EFL is to examine the key The growing global demand for English language learning and the
dimensions that contribute to enhancing language learning experiences limited opportunities for EFL students to practice their interactional
in universities. These dimensions represent essential features that need competence in traditional settings highlight the need for an in-depth
to be assessed and analyzed in order to provide a comprehensive and study of AI dialogue systems in university EFL contexts (Butler, 2019,
multifaceted understanding of the system’s effectiveness. Some re pp. 477–496; Camiciottoli & Campoy-Cubillo, 2018; Kawinkoonlasate,
searchers have referred to these dimensions as “evaluation criteria” or 2020). This is because AI dialogue systems can offer personalized,
“assessment indicators”, which help guide the examination of AI dia accessible, and controlled learning environments that facilitate mean
logue systems in education (Adenle, Chan, Sun, & Chau, 2020; Da Silva, ingful conversations and virtual immersion experiences (Tao et al.,
Fernandes, Limont, & Rauen, 2020; Gunnarsdóttir, Davidsdottir, Wor 2023). By focusing on interactions, researchers can better understand
rell, & Sigurgeirsdóttir, 2020). In this study, the term “dimensions” re and optimize the potential benefits of AI dialogue systems, such as
fers to facets of developments, trends, or issues in EFL that need to be tailoring input and difficulty, providing real-time feedback, and
assessed and analyzed, encompassing technical, theoretical, and peda enabling empirical research on EFL interactional practice (Divekar et al.,
gogical perspectives. The development of clear and specific research 2021, pp. 1–29).
questions can help to provide a more lucid understanding of these di Furthermore, it is essential to conduct an in-depth study on the
mensions. For example, one research question might be: implementation of AI dialogue systems for EFL, particularly on the
What are the key dimensions of AI dialogue systems for enhancing interaction strategies. This is important because interaction strategies
student’s EFL interactional competence? can help to (a) promote spontaneous initiation of engagement and
To address this question, it is helpful to draw on the findings of support EFL students’ wellbeing during their academic life at university
previous studies that have employed multidimensional evaluation (Ulum, 2020), (b) embed cross-cultural humor detection and response
frameworks. For instance, Kuhail, Alturki, Alramlawi, and Alhejori functions in the design of dialogue systems for EFL (Zhai & Wibowo,
(2023) utilized several dimensions, including educational field, plat 2022), (c) integrate cross-cultural awareness functions (Timpe-Laughlin
form, educational role, interaction style, design principles, empirical & Dombi, 2020), and (d) introduce cross-cultural empathetic functions
principles, and challenges and limitations, to evaluate an AI-based in EFL dialogue systems (Zhai & Wibowo, 2022). Zhai, Wibowo, and
chatbot for English language learning. Similarly, Zhai and Wibowo Cowling (2022), for example, explore the potential benefits, challenges,
(2022) proposed a framework for evaluating chatbots in language and methodologies associated with embedding humor and empathy into
learning that includes dimensions such as embedding cultural, empa language learning systems. Wang et al. (2022) highlight the impact of a
thetic, and humorous dimensions in chatbot design, pedagogical effec culturally-aware pedagogical agent on learners’ emotions, perceived
tiveness, and user satisfaction. Lastly, Li, Chang, and Wu (2020) used learning, and learning outcomes in computer-mediated communication.
multiple dimensions, such as technical quality, pedagogical effective It emphasizes the importance of considering cultural aspects in the
ness, and learner satisfaction, to evaluate an AI-based English dialogue design of pedagogical agents to enhance the learning experience for EFL
system. By synthesizing these findings, a comprehensive set of evalua students. The study concludes that a culturally-aware pedagogical agent
tion dimensions for AI dialogue systems in EFL can be identified. These can positively affect learners’ emotions, perceived learning, and
evaluation dimensions questions might include: learning outcomes. However, the importance of considering cultural
aspects and other interaction components, such as humor and empathy,
• Pedagogical effectiveness: How well does the AI dialogue system in AI dialogue systems for university EFL students has not been fully
support language learning outcomes? explored.
• Interaction style: How do students interact with the AI dialogue It is critical to investigate the role of the interaction components of
system, and how does this affect their learning experience? humor, empathy, and culture in AI dialogue systems for university EFL
• Design principles: What principles guide the development of the AI learners. This is because these interaction components can help to: (a)
dialogue system, and how do they contribute to its effectiveness? facilitate learner motivation, (b) support intercultural competence and
• Cultural, empathetic, and humorous dimensions: How well does the (c) address diverse learner needs. Huo (2022), for example, states that
AI dialogue system incorporate these elements, and what impact do there is a positive impact of cross-cultural awareness functions on EFL
they have on the learning experience? learners’ psychological well-being, academic involvement, and perse
• User satisfaction: To what extent are students satisfied with their verance. A holistic approach to EFL dialogue systems, including
experience using the AI dialogue system? cross-cultural humor recognition and empathic response features, con
• Technical quality: How well does the AI dialogue system perform in tributes to an effective and engaging learning environment (Zhai &
terms of accuracy, responsiveness, and other technical aspects? Wibowo, 2022). Zhang and Jing (2022) examine the use of AI technol
• Challenges and limitations: What challenges and limitations are ogies in facilitating cross-cultural communication and preserving
associated with the use of the AI dialogue system, and how can they intangible cultural heritage, demonstrating that AI technologies can
be addressed? bridge cultural gaps, translate languages, and foster an understanding of
diverse traditions and cultural practices. Incorporating these aspects in
By focusing on these evaluation dimensions and developing research EFL dialogue systems ensures that learners develop crucial intercultural
3
competence for effective communication in today’s globalized world. the fact that they seldom hear sounds and words spoken in English by
Inkster, Sarda, and Subramanian (2018) highlight the potential of people outside of the classroom. The authors believed that different
empathy-driven conversational AI agents like Wysa in promoting mental social and cultural elements might result in a variety of input and
well-being, suggesting that such agents can serve as a valuable digital communication sources for the first language. In this instance, the cul
tool for supporting users in their learning journey. By considering ture of students is characterized by a predominant use of the Javanese
emotional and cultural factors in the design of EFL dialogue systems, language in the students’ communication within the communities.
educators can better address the diverse needs of learners from various Because of this, students only get inputs in English sometimes outside of
cultural backgrounds. Collectively, these studies have shown the class. Zrekat and Al-Sohbani (2022) investigated interactional commu
importance of a holistic approach to EFL dialogue systems that considers nication issues among 74 Saudi EFL students and revealed inadequate
a range of emotional and cultural factors for creating an effective and practice of English in and out of the classroom as one of the variables
engaging learning environment. negatively impacting learners’ speaking skills. The participants in this
This paper consists of six sections. Section two discusses the meth study (44.07 percent strongly agreed and 55.93 percent agreed) stated
odology of reviewing, selecting and data analysis of articles on dialogue that there is a shortage of speaking practice both inside and outside of
systems. Section three presents an overview of the EFL dialogue system the classroom was identified as one of the variables negatively affecting
for interactional competence at the university level. This is followed by learners’ speaking abilities. In addition, the interaction is often
results of the findings on the interactional practice of dialogue systems hampered by large class sizes or by the use of the students’ native lan
in university education in Section four. Section five discusses the find guage (Jafari & Ansari, 2012). Thus, some scholars suggest that AI
ings and gaps of the selected studies. Finally, section six entails theo technologies, such as dialogue systems, are required to facilitate EFL
retical and practical implications, limitations and recommendations for university students’ expression and creativity (Bibauw, François, &
future studies. Desmet, 2019; Jafari & Ansari, 2012).
Dialogue systems for EFL employ the interactional approach and
1.2. Rationale of the chosen study provide a significant basis for EFL as conversation inherently provides
chances for input, output, and interaction. The dialogue system delivers
According to research that was released by The Brainy Insights, the input and adjusts the degree of difficulty which may be tailored to the
global market for English language learning is predicted to increase from level of university EFL students. It is common that EFL students are
USD 11.35 billion in 2021 to USD 35.78 billion by 2030, at a compound required to articulate their intended meaning at each multi-dialogue
annual growth rate (CAGR) of 13.6% throughout the period of fore turn which can be seen as a forced output (Bibauw et al., 2019). In
casting that spans 2022–2030 (Insights, 2021). According to the British classroom settings, Dialogue systems for EFL provide students the
Council (Wright, 2022), there are 750 million people who speak English chance to practice meaningful conversations in a “virtual immersion”
as a second or additional language. It is anticipated that the number of environment, which not necessarily as effective as having a conversation
people studying English in different parts of the globe will continue to in the target language with a native speaker, but the opportunities can
rise. According to the research titled “The English Effect” published by offer equivalent characteristics. University learners can spontaneously
the British Council, there are presently 1.75 billion individuals world get access to the whenever they intent to, as the system does not lose its
wide who speak English language. This is one out of every four persons patience when faced with a speaker who is having difficulty (Bibauw
on the globe. It is predicted that by the year 2020, there will be 2 billion et al., 2019). The dialogue systems also provide a learning environment
individuals utilizing the language. English as a Foreign Language (EFL) that can be fully controlled and is potentially configurable towards
is becoming a working language for a growing number of influential optimal conditions on all impacting factors (feedback, learner modelling
people throughout the globe since it is necessary for the study of a wide and adaptivity, and motivational support). This environment can be
variety of courses at universities. However, existing studies found that used not only for learning but also for research purposes. Dialogue
university students in many EFL settings do not have many opportunities systems for EFL have the potential to give completely monitored settings
available to enhance their interactional competence outside of the for conducting empirical research on EFL interactional practice as it
classroom or even inside the classroom, as the spoken interactional eliminates the variable and unpredictable nature of a human interloc
practice is frequently limited to interactions between the teacher and utor (Bibauw et al., 2019).
classmates (Wahyuningsih & Afandi, 2020; Xie, 2017; Zrekat &
Al-Sohbani, 2022). Following concerns from stakeholders, the current 2. Systematic review method
teaching of spoken English and interactional competence is still viewed
as unsatisfactory. This is due to the fact that many graduates still lack the This section conforms to the recommended items reported in the
ability to communicate in a foreign language, which prevents them from systematic review and provides insight into the applications of the
participating fully in transcultural and global society (Beshir & Yigzaw, dialogue system for EFL acquisition. A systematic review has been
2022; Darmajanti, 2017). characterized as a form of knowledge synthesis that outlines exploratory
University students in many EFL settings do not have many oppor research questions by methodically selecting, identifying, and synthe
tunities available to use the target language outside of the classroom or sizing existing information (Colquhoun et al., 2014). A systematic re
even inside the classroom, as the spoken interactive practice is view is useful for mapping the landscape of published literature and
frequently limited to interactions between the teacher and classmates. identifying research gaps, defining research objectives, and providing
Xie (2017) used participant observation and questionnaires with rating policymakers with suggestions (Tricco et al., 2018). This study aims to
and open-ended questions as research tools to study 128 EFL students understand the advantages of a dialogue system for EFL learning to
from various academic disciplines taking business English classes. The enhance long-term learning outcomes and promote university ESL stu
findings indicate that approximately 24% of the university students re dents’ well-being. To reach this goal, this study employs Wolfswinkel
ported that their limited English communicative skills prevented them et al.’s (2013) five-step method for performing a systematic literature
from making full use of their target language with teachers and class review and analysis. This five-step approach allows researchers to
mates in class. On the other hand, only 1.6% of the participants reported conduct a comprehensive search, assess and analyze the articles
using English outside of the classroom between 80% and 100% of the collected from databases. The five-step method entails (a) determining
time. According to the percentages, the amount of time spent by par the scope of a review, (b) conducting a literature search, (c) choosing the
ticipants using their target language both inside and outside class is still final samples, (d) evaluating the samples using content analysis, and (e)
insufficient. Wahyuningsih and Afandi (2020) found that universal presenting the study results.
students’ progress in English, particularly in speaking, was hindered by
4
2.1. Determining the scope of a review (b) Does the paper discuss supporting learning activities used in
dialogue systems aimed to improve EFL students’ performance of
This initial stage involves the clarification of the inclusion and interactional competence during university education?
exclusion criteria for relevant sources, as well as the criteria for identi (c) Does the paper discuss EFL students’ engagement with dialogue
fying and retrieving relevant sources from the literature. To be included systems aimed to enhance EFL students’ performance of inter
in the review, the journal articles needed to focus on the following se actional competence during university education?
lection criteria: (a) published in English in a full-text article, (b) asso (d) Does the paper discuss learning outcomes of EFL students’
ciated with dialogue system with EFL in university, (c) focused on engagement with dialogue systems focused the performance of
communicative or interactional competence (d) provided empirical interactional competence on during university education?
data, and (e) published between January 2013 and August 2022. We (e) Does the paper discuss EFL students’ perceptions and attitudes
selected the publication range between January 2013 and August 2022 towards dialogue systems focused the performance of interac
due to the: (a) rapid development of AI and natural language processing tional competence on during university education?
(NLP), particularly with the emergence of deep learning and (f) Does the paper discuss challenges and problems of designing
transformer-based models, (b) increasing adoption of AI dialogue sys dialogue systems for EFL in relation to enhancing interactional
tems for EFL education, and (c) evolving pedagogical approaches and competence during university education?
global demand for English language learning which have prompted
educators and researchers to explore innovative pedagogical The reason for discussing the technologies used for dialogue systems,
approaches. students’ engagement, EFL students’ perceptions and attitudes is
Articles are not considered when they (a) did not concentrate on because these technologies can play a significant role in increasing
dialogue systems for EFL in university, (b) focused on other language students’ level of interactional competence, and subsequently
learning, (c) were written in languages other than English, (d) were improving their EFL students’ performance during their university ed
editorials, (d) were opinion pieces, and (f) were dissertations. There are ucation (Alsadoon, 2021; El Shazly, 2021; Hsu et al., 2021).
several reasons for excluding editorials, opinion pieces, and disserta
tions. First, the purpose of this systematic review is to provide a 2.2. Conducting a literature search
comprehensive, unbiased, and objective analysis of the available liter
ature on a specific topic. Firmstone (2019) claims that systematic re The second stage involves entails executing the search query against
views need to focus on empirical research studies that follow a the chosen databases in order to get the search results. Several of da
well-defined methodology and adhere to established research stan tabases including ProQuest, IEEE, Science Direct, and Web of Science
dards. Second, editorials and opinion pieces often represent the sub were used for the review process. These databases were chosen based on
jective perspectives and opinions of the authors (Firmstone, 2019). their importance to education (e.g., Science Direct) and AI (e.g., IEEE) in
Third, Siddaway, Wood, & Hedges (2019) pointed out that editorials and order to assure the review’s comprehensiveness. Google Scholar was
opinion pieces generally lack the methodological rigor and detailed selected to search the database simultaneously to expand our retrieval of
reporting of methods and results found in research articles, making it data and Web of Science was included as a secondary database to
difficult to assess their quality and relevance to the review. This is identify additional material that may be relevant to our investigation.
supported by Firmstone (2019) who believes that the use of editorials, An initial search was carried out by utilizing four sets of searches and
opinion pieces, and dissertations may introduce bias and potentially each search terms is comprised of several variations of the term English
skew the overall findings of the review. In addition, Thomas & Skinner, as a foreign language as a medium of instruction, in which includes
2012 state that even though dissertations can contain valuable research, English as a foreign language (EFL), teaching English as foreign lan
they are often not as extensively peer-reviewed as journal articles. The guage (TEFL), teaching English to speakers of other languages (TESOL).
use of dissertations for conducting reviews may also increase the risk of The second set of search phrases focuses on concepts associated with
including studies with methodological weaknesses or biases. Table 1 interactional competence in which includes interactional skills, inter
summarizes and presents the inclusion and exclusion criteria. actional tasks, interactional competence, conversational English,
The following criteria were used to select the articles from the da speaking skills. The third set of search phrases consists of terms associ
tabases for this study: ated with AI in which includes AI, natural language processing (NLP),
neural network (NN), intelligent tutoring system (ITS), machine
(a) Does the paper discuss technologies used for dialogue systems learning (ML), machine intelligence (MI), deep learning (DL), and
aimed to enhance EFL students’ performance of interactional classical algorithm output (CAO). The fourth set of search relates to the
competence during university education? term of university education in which includes associate program, un
dergraduate program, academic program, diploma, further college
program, postgraduate program, graduate diploma, master, and philo
Table 1 sophical degree. However, when we combined these categories and put
Inclusion and exclusion.
terms into the search bar on ProQuest, IEEE, Science Direct, Web of
Inclusion Exclusion Science, and Google search engine, zero data was retrieved. Thus, we
(a) Articles must be published in (a) Articles did not concentrate on have only included search terms such as EFL, interactional competence,
English in a full-text article, Dialogue systems for EFL in AI and university education. During the course of the search, a total of
(b) Articles must be associated with university, 211 publications were retrieved where they have an abstract or title that
dialogue system and EFL in (b) Articles focused on other language
university, learning,
included all four sets of search terms.
(c) Articles must focus on (c) Articles were written in languages There are several steps involved in the review process. Firstly, the
communicative or interactional other than English, authors restricted the publication period to the past ten years, from
practice, (d) Articles were editorials on Dialogue January 2013 to August 2022, to ensure that the literature is relevant
(d) Articles must provide empirical systems for EFL in universities,
(Bener, 2021). Then, duplicate copies of the research were removed, and
data, and (e) Articles were opinion pieces on
(e) Articles have to be published Dialogue systems for EFL in this brought the total number of studies down to 180. Next, both
between January 2013 and August universities, and peer-review journals and conference proceedings were included. In
2022. (f) Articles were dissertations on accordance with this criterion, the search did not include trade publi
Dialogue systems for EFL in cations, editorials, books, or review articles. The search was further
university.
narrowed down to only include articles that were written in English
5
pertaining to EFL, and this brought the total number of results down to by AI dialogue system for EFL learning. Initial coding, often known as
115 papers. Following this, the titles and abstracts of the selected pub “open coding,” refers to the initial phase in the coding process, espe
lications were analyzed to ensure whether or not they were pertinent to cially for qualitative methodologies such as grounded theory. During
university students’ interactional competence that made use of AI this first phase, qualitative data is divided down into distinct fragments
technology. According to this criterion, we excluded studies that and assigned codes (Vollstedt & Rezat, 2019b). Open coding refers to the
involved non-EFL learning, e.g., focus on Turkish language learning by first interpretative procedure through which raw research data are
Guzey, Yildiz, Demir, and Aksu-Dunya (2022) non-communicative classified and systematically evaluated. In grounded theory research,
competence, e.g., examine the difference between media language open coding was presented as a crucial methodological instrument for
learning and traditional language learning by Tripathi, Ray, Sinha, and qualitative data analysis. (Vollstedt & Rezat, 2019b). Axial coding is a
Ahmed (2021), non-university level education, e.g., such as research qualitative research approach that entails connecting data in order to
carried out in senior high school by Nirwana and Suhono (2022), thesis identify codes, categories, and subcategories grounded in the voices of
writing, e.g., investigate learners’ learning preference in class by Arung, participants within gathered data. In other words, axial coding is one
Rafli, and Dewanti (2019), not student learners, e.g., focus on teachers’ method for constructing data links (Vollstedt & Rezat, 2019b). Selective
attitudes towards using computer assisted language learning by Gorjian coding refers to the procedure of selecting one category as the core
(2017) or the use of non-AI technologies. e.g., the use of a blackboard category and associating all other categories with it. The fundamental
program by Al-Oqaily, Salam, and Na (2022)). This resulted in 28 concept is to construct a singular plot around which everything else
research being selected for the final round of consideration. Table 2 il revolves. It is believed that such a fundamental notion always exists
lustrates the criteria for inclusion and exclusion along with the number (Vollstedt & Rezat, 2019b).
of studies that were retained after determining whether or not the Initial codes were generated through open coding, the authors
criteria were met. refined the initial codes through iterative processes, including compar
ison, group discussion, and further code expansion. Along the process,
2.3. Choosing the final samples initial coding schemes for AI dialogue systems for EFL were identified,
such as technological integrations, task design, students’ engagement,
The third stage consists of selecting the final samples so that a full and learning objectives, as more articles were analyzed, they realized
analysis may be performed on them. The search is confined to the titles that the dialogue systems for EFL required further differentiation. These
and abstracts so that researchers can concentrate on the search results. coding schemes allowed for a more comprehensive understanding of the
All original article titles and abstracts are scrutinised for their relevance various aspects and dimensions of EFL learning supported by AI dia
to AI dialogue systems which are used to enhance EFL university stu logue systems (Vollstedt & Rezat, 2019b).
dents’ interactional competence. This resulted in the discovery of 211
related articles. Then, duplicate articles are eliminated. There are 28 (1) Technological Integrations: This code highlights the different
articles remaining for further evaluation. technological components and tools that can be integrated into AI
dialogue systems for EFL to enhance learning experiences, such as
2.4. Evaluating the samples using content analysis virtual reality, multimedia, and game-based learning
environments.
Based on the research objectives, a total of 28 articles were selected (2) Task Design: This code emphasizes the importance of designing
for further examination and evaluation. In this study, a systematic re engaging, meaningful, and contextually relevant tasks for EFL
view was conducted in which data were gathered, processed, identified, learners within AI dialogue systems to promote language devel
and summarized. A six-step procedure was employed to determine the opment and authentic language use.
recurring dimensions. The first phase involves conducting a thematic (3) Students’ Engagement: This code focuses on the level of interest,
analysis of the data to acquire a more in-depth understanding. Then, the motivation, and involvement that AI dialogue systems can
initial codes were created. The third and fourth phases entail identifying generate among EFL learners, which is crucial for successful
sub-dimensions and reviewing the sub-dimensions. The fifth phase language acquisition, and
consists of compiling all of the pertinent concepts. Finally, the data were (4) Learning Objectives: This code addresses the specific goals and
analyzed to ensure relevance to the study’s objectives. To ensure rele outcomes that AI dialogue systems for EFL aim to achieve, such as
vance and review quality, the literature that emerged from the databases improving vocabulary, grammar, pronunciation, listening,
has been screened twice based on the PRISMA process. PRISMA stands speaking, reading, and writing skills.
for Preferred Reporting Items for Systematic Reviews and Meta-
Analyses. It is a minimal set of criteria for reporting in systematic re As the authors analyzed more articles, they also encountered some
views and meta-analyses based on evidence. Fig. 2 illustrates that the potential limitations and disagreements in the literature. For instance,
PRISMA statement contains a 27-item checklist and a four-phase (1) Technological Limitations: This code represents the challenges and
diagram. constraints faced by AI dialogue systems for EFL, such as imperfect
natural language understanding, limited cultural awareness, or acces
2.4.1. Coding scheme sibility issues for some learners, and (2) Novelty Effect: This code cap
In order to emphasize on the interactional competence supported by tures the idea that the initial enthusiasm and engagement with AI
AI dialogue systems for EFL at universal level, a mixed method dialogue systems for EFL may be driven by the novelty of the technology
combining both deductive or “top-down” and inductive or “bottom-up” rather than its inherent educational value. Over time, this effect may
approaches proposed by Elo and Kyngäs (2008) was adopted into our diminish, causing a decrease in students’ engagement and learning
coding scheme. These methods have had significant application across outcomes.
many different fields of study, including human-interaction dialogue In the second stage of the research process, axial coding was
systems (Diederich, Brendel, Morana, & Kolbe, 2022), and the whole employed by the authors as a systematic means of analyzing the findings
process of coding was carried out in Microsoft Word and Microsoft Excel. derived from the initial open coding stage (Vollstedt & Rezat, 2019a).
The three-step coding method proposed by (Vollstedt & Rezat, 2019b) During this phase, each author independently examined the emergent
was followed by the authors, who carefully read all of the transcripts and codes, focusing on identifying relationships between them and orga
identifying dimensions of the retrieved articles until saturation was nizing the data into meaningful categories and subcategories (Harati,
achieved. The three-step coding entails: initial coding, axial coding and Nooshinfard, Isfandyari-Moghaddam, Babalhavaeji, & Hariri, 2019). To
selective coding are employed in identifying six dimensions supported ensure the rigor and reliability of the analysis, the authors engaged in
6
C. Zhai and S. Wibowo
Table 2
General profile of the reviewed studies on AI dialogue systems.
No. Authors and Year Focus of Study Research Design Participants Technological Task Design Students’ Learning objectives Technological Novelty
Integration Engagement Limitation Effect
1. Ayedoun et al. To foster students’ Survey 40 university DMS • Dialogue Boost students’ EFL students’ engagement Unnatural sound N/A
(2019) readiness towards EFL undergraduate interactions motivation for is enhanced towards Gibberish response
verbal interactions and graduates • Negotiation for communicating in communication
from Japan meaning EFL
2. Ayedoun, Hayashi, To enhance EFL students’ Questionnaire 60 Japanese DMS • Negotiation for Enhance students’ The dialogue system N/A N/A
and Seta (2020) wiliness to converse in the university students meaning participation in motivates EFL students’
target language dialogue willingness to
communicate
3. Chen, Yang, and Lai To investigate students’ Survey 29 university ASR & IPA • Conversation Encourage students A helpful tool to improve Difficult to N/A
(2020) perceptions toward the use students from to speak with the students’ pronunciation understand heavy
of IPA Taiwan dialogue system and vocabulary accented speech
4. Divekar et al. (2021) To research immersive A mixed-method 10 university • ASR • Cultural Enhance speaking Students’ vocabulary Technology bugs Yes
technologies and students from • SDS knowledge and writing retainment and
collaborative learning in a China • CILLE • Role play conversation skills
dialogue system • Negotiation for significantly improved
meaning Less anxious
5. Dizon (2017) To examine whether IPA, Interview Single 4 s-year university • ASR • Storytelling Students enjoyed EFL students regards IPA Odd robotic voices N/A
Alexa, is able to group students from • Audiolingual • Question & talking with Alex as a useful tool to enhance
understand EFL English Japan answer language learning
utterances • Interview
Earplay
6. Dizon (2020) To investigate speaking Quasi- 37 first- and ASR • Storytelling Engage students to A useful tool to support Unable to provide N/A
development, such as experimental second-year • Question & talk for more than listening comprehension modifications
conversation, discussion, design Single university EFL answer 20 min and speaking proficiency
and presentation skills group students from • Negotiation for
7
Japan meaning
7. Dizon and Tang Use a dialogue system, Qualitative 2 fourth-year IPA Conversation Enjoy talking with Dialogue systems N/A N/A
(2019) Alexa, for self-study method Single university students Alex facilitate pronunciation
group from Japan
8. El Shazly (2021) To examine the role of AI Quasi- 48 students from • ASR Role-play Slightly enhanced AI dialogue systems N/A Yes
in interactional practices experimental Egypt • Audiolingual students’ speaking improved interaction and
Single group abilities oral communication
But increased instead
reducing students’
anxiety

9. Hsu et al. (2021) To explore the effects of Quasi- 40 s or third year IPA • Read aloud Students tend to IPAs foster the students’ Without visual N/A
Amazon Echo Show on EFL experimental of university from • Describe photos interact more with speaking abilities due to displays
students’ listening and Single group Taiwan • Answer Echo highly interactive
speaking skills questions learning environment.
Less stressful
10. Hsu, Chen and Yu To offer a chatbot system, Pre- and post- 100 students from Task-oriented Conversation Increase students’ The findings indicate that N/A N/A
(2021) TPBOT, for EFL students to experiment Taiwan chat robot engagement with students are happy with
reduce their fear of Control & system TOBOT this TPBOT and think that
speaking English. experiment it has helped them to
groups enhance their English-
speaking abilities.
11. Johnson (2019) To present two case studies Survey 209 university D3 Conversation Improve English Enskill®English is useful Dialogue choices N/A
in the data-driven from Serbia and proficiency to suit for acquiring spoken are too limited and
development of Croatia the demands of more English abilities. restricted, and
Enskill®English for foreign advanced students encountered bugs.
language learning
12. Kim (2016) To develop an English- Survey 181 college Speech to text Voice chat Engage students in All of the participants N/A N/A
speaking system that students in Korea various voice chats showed statistically
(continued on next page)
Table 2 (continued )
improves students’ significant improvements

speaking abilities in their speaking abilities.
13. Kim (2017) To evaluate voice-based Survey Between 123 freshman Audiolingual • Questions and Encourage students Dialogue systems play an Still use pre- N/A
dialogue systems on EFL group university students answer in meaningful important role in voice- determined topics
students’ negotiation for from Korea • Negotiation for dialogue based interaction using and pre-stored
meaning meaning meaning negotiation phrases
approaches
14. Kim, Cha, and Kim To examine how AI Pre- and post- 49 university N/A technology Asking and Motivate and guide The students’ ability to N/A N/A
(2021) chatbots affect students’ experiment students integration. Only answering short students’ speaking speak was improved,
speaking skills, motivate names questions or skills especially their
and shape students’ Replika, Andy, opinions pronunciation,
speaking experiences Google Assistant intonation, and stress.
15. Li and Peng (2021) To evaluate a blended A mixed-method 59 freshman Audiolingual Conversation Positive attitudes Students EFL learning N/A N/A
teaching mode based on an Single group university students can be found interests and attitudes
AI learning platform from China towards AI learning towards AI dialogue
platform systems are enhanced
Anxiety was reduced.
16. Li et al. (2020) To examine the feasibility Survey 28 university TTS Conversation Continuous student Positive perceptions of Some students had N/A
of replacing the students engagement using AI the system’s user issues with their
conventional English- dialogue systems interface, learning style, speech not being
speaking method with the and efficacy were properly
dialogue system recorded. recognized.
17. Mazzilli (2021) To investigate how Questionnaire 28 third year CALL Conversation Improvement of The use of chatbots to There were N/A
students perceive the university student conflict resolution enhance conflict- coherence issues
chatbot Elbot and it affects learning German abilities resolution skills in EFL during interaction
the students’ behavior in with the chatbot.
8
conflict-resolution skills
18. Mohammadzadeh Through AI dialogue Quasi- 45 students from • ASR Conversation Engage in problem- Dialogue system enhances N/A N/A
and Sarkhosh (2018) system to improve experimental Turkey • SDS solving activities students’ performance in
speaking ability Between group speaking
19. Morino et al. (2017) To investigate the effect of Single group 120 university • ASR • Storytelling Enjoy storytelling The course increased Incapacity to have N/A
digital storytelling on students from • SDS • Negotiation for conversations students’ awareness of lengthy
Japanese EFL students’ Japan meaning CALL and improved their conversations
attitudes and awareness speaking proficiency.
toward CALL
20. Moussalli and To investigate API’s ability Quasi- 18 university IPA • Role-playing Encourage students Some improvements were N/A N/A

Cardoso (2021) to increase EFL students’ experimental students from to revise and self- reported on phonological
phonetic awareness and Single group various correct awareness but no
allomorph production backgrounds significant difference
between pre-and post-
tests.
21. Najima, Kato, To encourage students to Interview Single 10 university • ASR • Conversation Enjoy repeating The study demonstrated N/A N/A
Tamura, and improve speaking group students from • SDS • Question & speaking tasks the effectiveness of the
Yamamoto (2021) proficiency Japan answer system for remote
• Negotiation for learning and
meaning concentration level.
22. Tegos, Demetriadis, To use MentorChat to Post-task 30 Russian ITS • Conversation Engage students in The ‘weak’-directed agent N/A N/A
and Tsiatsos (2014) trigger constructive Questionnaire university student Collective conversations intervention mode seems
interaction amongst Focus group studying English discussion to foster a more
students constructive discourse
among students.
23. Tai (2022) To examine the influence Interview 89 college IPA • Oral proficiency Students provided The findings indicate that N/A N/A
of IPAs on English as a freshmen from tests high-quality oral the usage of an IPA
foreign language students’ China • Interview input outside of class
(continued on next page)
Table 2 (continued )
spoken competence significantly enhanced

outside the classroom. the oral competency of
EFL students
24. Timpe-Laughlin, To examine the utility of Survey Single 47 EFL tertiary SDS • Conversation Verbal, gestural, and SDS tasks are effective for Speaking too fast N/A
Sydorenko, and spoken dialogue systems group level students in • Question & physical interaction practicing speaking and or interrupting
Dombi (2022) (SDSs) for EFL learning the USA answer collecting speaking
• Negotiation for performance data
meaning
25. Wu et al. (2020) To identify emotional Interviews 32 participants • IPA • Question & Inclined to EFL speakers prefer visual Difficult to N/A
issues in EFL IPA user Between group from a European • Audiolingual answer participate in well- feedback to support their recognize students’
experience university • Negotiation for orchestrated interaction which boosts utterances
meaning speaking practice their confidence
26. Xie, Liu, Chen, and To incorporate a Survey Between 40 university RAI & GAI • Multiturn Bi-directional A dialogue system that is Lack of seamless N/A
Liu (2021) motivational online group students from conversation communication embedded in MI to integration affects
9
conversational agent China improve student students’

(MOCA) to enhance engagement in experience
learner engagement in collaboration tasks
computer-supported
collaborative learning
(CSCL)
27. Yin and Satar (2020) To explore the frequency Between group 8 undergraduate CALL Negotiation for Engage in Low-level students benefit Not able to Yes
and patterns of NfM university students meaning negotiation for the most, and higher-level understand Out-of-
language learning from China meaning students are dissatisfied context response
with the systems.

28. Zou, Liviero, Hao, To use an AI dialogue A mixed-method 113 Y1 and Y2 CALL • Pronunciation Students preferred Positive attitudes toward Poor voice N/A
and Wei (2020) system to support students’ university students • Task-based AI as lack of teacher the AI dialogue systems recognition
speaking skills in EAP from China conversation feedback for speaking skills Limited prosodic
courses • Presentation system
discussions with two independent specialists who brought their own Third, we utilized the integrated core viewpoint to explain the patterns
expertise and perspectives to the interpretation of the data (Böhm, 2004, and interactions between the phenomenon’s aspects by use of an
p. 270). By incorporating these external viewpoints, the authors aimed explanatory matrix, thereby disclosing everything involved. This is not
to minimize potential biases and enhance the credibility of the research an exhaustive list, but it does include the majority of technologies that
findings (Lianto, 2019). The collaborative approach adopted during the are commonly and often utilized in EFL educational research dealing
axial coding stage facilitated a comprehensive and nuanced under with interactional competence. This paper identifies 6 dimensions out of
standing of the subject matter. Through ongoing dialogue and negotia 28 studies, such as: technological integrations (n = 28), task design (n =
tion, the authors and the independent specialists were able to 28), students’ engagement (n = 28), learning objectives (n = 28),
collectively refine the categories and subcategories, ultimately reaching technological limitation (n = 10) and novelty effects (n = 3).
a consensus on the interpretation of the data (Harati et al., 2019). This Technology integration in language learning involves various sub-
iterative process not only contributed to the development of a robust dimensions, such as automatic speech recognition (ASR), audiolingual,
analytical framework but also fostered a sense of shared ownership over intellectual personal assistant (IPA), spoken dialogue system (SDS),
the research findings (Vollstedt & Rezat, 2019a). computer-assisted language learning (CALL), dialogue management
In the third stage of the research process, the authors engaged in systems (DMS), cognitive immersive language learning environment
selective coding with the aim of clustering the axial codes gathered (CILLE), data-driven development (D3), intelligent tutoring system
during the previous stage. This step involved a thorough examination (ITS), and retrieval-based AI & generative AI (RAI & GAI). In designing
and synthesis of the axial codes to identify key dimensions, sub language learning tasks, aspects include conversation, negotiation for
dimensions and establish relationships between them, ultimately lead meaning, question and answer, and storytelling. Aspects of engagement
ing to the development of a cohesive framework that captured the of students includes bi-directional communication and productive lan
essence of the findings. To maintain the rigor and validity of the guage exercises. Learning objectives include academic achievement,
research, the authors carried out selective coding individually, ensuring increased concentration, enhanced cultural awareness, and learning
that each researcher’s unique perspective and insights were taken into through collaboration. Aspects of technological limitations include un
account (Vollstedt & Rezat, 2019a). Throughout this process, the au natural voice, incomplete phrases, failed communication, and gibberish
thors held regular discussions with subject matter experts to review their outputs. Additionally, technology novelty can lead to a novelty effect.
progress, share interpretations, and address any discrepancies or un Fig. 1 illustrates a hierarchical diagram for enhancing EFL university
certainties that arose. These collaborative efforts were vital in ensuring students’ interactional competence.
that the analysis was comprehensive, well-founded, and grounded in the This paper aims to present an overview of the technological appli
data (Holton, 2007). The continuous exchange of ideas and the inte cations of dialogue systems for EFL in university education. The research
gration of expert feedback contributed to a more nuanced understanding articles retrieved from the databases were screened twice. First, an ex
of the subject matter over the research findings (Moghaddam, 2006). amination of the abstracts and titles of the retrieved material is carried
This iterative process continued until the authors and the subject matter out in order to determine whether these articles match the minimal
experts reached a point of complete accord, signifying that a shared inclusion criteria. Second, the entire text of the publications that were
understanding and agreement had been achieved regarding the central included in the study was analyzed and retrieved using the CQUniversity
themes, relationships between codes, and overall framework derived database retrieval tool. The databases yielded a total of 211 articles.
from the data (Vollstedt & Rezat, 2019a). By reaching this stage, the After screening and removing duplicates and unsuitable content, the
researchers ensured that the analysis accurately reflected the insights total number of articles was reduced to 180 articles. Then, after the
and patterns emerging from the collected data, ultimately enhancing the eligibility screening procedure, a total of 115 articles were retrieved and
credibility and generalizability of their findings. assessed for eligibility. Then, any remaining publications that were not
After three steps of open coding, axial coding and selective coding, directly relevant to AI dialogue systems for EFL or EFL learning were
we adopted dimensional analysis to identify and categorize dimensions excluded from this research. As a result, a total of 28 articles were
related to interactional competence supported by AI dialogue systems chosen for this study.
for EFL at a university level. Dimensional analysis is a technique for
characterizing complex phenomena in which meaning is socially 3. Overview of EFL dialogue system for interactional
established based on perspective and context according to grounded competence
theory methodology and social interactionism (Morse et al., 2016). The
purpose of dimensional analysis is to construct theory based on data Early in the 1980s, academics were motivated by the insights learned
analysis, including discovering and characterizing important di from previous computer-assisted dialogue system development attempts
mensions, inferring correlations, and establishing linkages between di to define new goals for dialogue systems (Tai, 2022). Underwood (1984)
mensions (Bowers & Schatzman, 2016). was the first person to advocate for the creation of communicative
dialogue systems, a strategy that would use recent advances in AI to
2.4.2. Method of dimensional analysis integrate meaningful communication practice in tutorial systems.
We followed Kools et al.’s (1996) influential three steps when apply Among other variables, this would be performed via verbal connection
dimensional analysis. First, we generated or identified dimensions and with others. FAMILIA, which was developed by Underwood in 1982, is a
their qualities, and then sub-categorized them to expose tentative no written Spanish conversational program. Its pattern-matching capabil
tions by expanding data. This portion of the preliminary study focuses ities were somewhat inspired by the first chatbot, ELIZA (Underwood,
on identification, since codes take the shape of AI dialogue systems for 1984).
EFL. This procedure continues until a critical mass of dimensions and Over the last decade, technological advancements in dialogue sys
attributes has been attained. Second, we created an explanatory matrix tems for EFL seem to have merged into new systems that employ so
and assigned higher weight to certain aspects while disregarding others phisticated NLP in tandem with more clear learning goals and better
(similar to constant comparison in Grounded Theory). Each dimension strategic evaluation techniques (Tai, 2022). Focusing on corrective
was elevated to this level during the process of determining the central feedback, dialogue systems like Sasha involve sophisticated conversa
perspective, and the dimension that provides the greatest explanation tion management and extremely granular evaluations of effectiveness.
for the relationship between dimensions is ultimately selected as the Multimodality, relevant and authentic communication settings,
central or key perspective from which to organize the data (Kools, goal-oriented interactions, mixed initiative, and complex conversation
McCarthy, Durham, & Robrecht, 1996). The remaining aspects are management are already partially shown by these systems, which will be
classified as salient, relevant, marginal, or irrelevant (Kools et al., 1996). crucial for future enhancements to the EFL dialogue system (Petersen,
10
Fig. 1. A hierarchical diagram for enhancing EFL university students’ interactional competence.
2010). conversational turns co-construct meaning in dialogue systems. Without

Recently, large-scale pre-trained models (PTMs), bidirectional a set of turns, the systems cannot be considered as dialogue-based ones.
encoder representations from transformers (BERT), and generative pre- Kramsch (1986) defines the term “interaction” as a communicative ex
trained transformers (GPT) have recently garnered enormous success change that entails negotiating intended meanings, and modifying one’s
and established themselves as an important milestone in the field of AI. speech to the impact one aims to have on the listener. It requires
Because of their intricate pre-training objectives and massive model defining one’s own objectives, anticipating the listener’s reaction and
parameters, large-scale PTMs are able to effectively extract information probable misunderstandings, and getting at the closest possible obser
from both labelled and unlabeled data sets. PTMs are autoregressive vation between intended, perceived, and expected meanings. Thus,
language models that, with the use of deep learning, generate writing interactional competence may be defined as a theory of the knowledge
that is eerily similar to that produced by humans; these models have that participants bring to and actualize in interaction, and includes an
approximately 175 billion parameters. explanation of how this information is gained (Young, 2011).
The majority of dialogue systems for EFL is item-based, and the
systems focus on the discussion as the instructional unit. This distin 4. Results
guishes it from production activities centered around separate items,
which is similar to a sentence found in most language instructional Through open coding, axial coding, and selective coding, we deter
courseware (Heift & Schulze, 2015). A conversation, as opposed to being mined that it would be appropriate to classify and code the dimensions
a syntactic unit, is a pragmatic unit, containing interactional practice that were identified in the chosen studies into a six-dimension classifi
and numerous complex phenomena that language students are required cation of EFL dialogue systems, each of which has its own objective and
to gain communicative competence (Heift & Schulze, 2015). Multiple function in optimizing or limiting learners’ interactional competence in
11
Fig. 2. The PRISMA flowchart.
university level of study. Each dimension includes multiple primary explaining the language forms (Hismanoglu & Hismanoglu, 2011).
families of AI approaches. This is by no means a complete list, but it does Learner engagement is a measure of the quantity and quality of a
include the majority of AI approaches that are commonly dealing with learner’s involvement in their learning (Wu et al., 2020). Learning ob
learners’ interactional competence in university level of study. Table 2 jectives are statements that clarify what students are expected to acquire
presents six dimensions, including (1) technological integration, (2) task when learners have accomplished a course or programme (Morino,
design, (3) student engagement, (4) learning objectives, (5) technolog Lopez, & Ono, 2017, pp. 367–373). A technological limitation is the
ical limitations, and (6) the novelty effect. inability of computer software or hardware to perform a certain function
Technological integration in language learning is the use of tech (McTear, Callejas, & Griol, 2016). The novelty effect refers to the pro
nology to improve students’ the learning experience (El Shazly, 2021). pensity for performance to initially improve when new technology is
Technology enhances students’ opportunities for authentic connection implemented—not because learning or accomplishment have really
with native speakers and other language learners of varying proficiency improved, but rather as a result of increased interest in the new tech
levels, both within and outside of the classroom (El Shazly, 2021). Task nology (Fryer, Ainley, Thompson, Gibson, & Sherlock, 2017).
design is a systematic way in which task description, workflow and task The finding indicates that AI dialogue systems serve a communica
execution plan are structured. Task design for language acquisition tive purpose of EFL acquisition in university education as students find
helps to facilitatee language learning by emphasizing on the regular dialogue systems alleviate their speaking anxiety, enhance their vo
communication and the usefulness of the language rather than cabulary retainment and offer immediate feedback to boost their
12
willingness to interact in the target language. However, one paper finds pronunciation and vocabulary (Ayedoun, Hayashi, & Seta, 2019).
that AI dialogue systems increase students’ anxiety instead of reducing Additionally, the processes of dialogue systems can alleviate students’
it. speaking anxiety (Bashori et al., 2020). Via the audiolingual method,
The target language in the dialogue system studies primarily focused dialogue systems identify and adapt individual students’ interactive
on learning EFL. As for research design, most studies employed single- behaviors to successfully engage students in the target language use
group experiments (n = 10), followed by survey (n = 6), between- (Hanif, 2016). This theoretical viewpoint in social psychology demon
groups (n = 5), a mixed-method (n = 3), and questionnaire (n = 3) strates a similar common basis in learning (Bashori et al., 2020). In order
(see Table 2). With regards to task designs, most studies employed to apply this approach to dialogue systems, it is vital for dialogue sys
conversation strategy (n = 12), followed by negotiation for meaning tems to exhibit similar dimensions and knowledge. This increased lan
strategy (n = 10), questions & answers strategies (n = 6), and pre and guage engagement, enhanced speaking abilities, and increased desire
post-experiment design (n = 2) (see Fig. 3). A single-group experiment is and interest in learning. Nevertheless, novelty effects have been recor
a type of quasi-experiment in which the result of interest is assessed only ded as hierarchically structured activity designs of dialogue systems are
once after the exposure of a non-random set of participants to a inclined to leave little variation for students to respond impromptu
particular intervention. The purpose of this experiment is to measure the (Fryer et al., 2017). Table 2 contains information on the articles that
impact of the intervention, such as to evaluate a training program (Ryu were chosen, such as the names of the authors, the focus of the studies,
& Cheong, 2017). A between-group study design is an experiment that research design, participants in the study, implantation duration, tech
employs two or more groups of participants, each of whom is put nological integration, task design, learning objectives, and technical
through a different testing process simultaneously (Ryu & Cheong, limitations.
2017). A pre- and post-experiment design is a method of research in The second technique basis was applying automatic speech recog
which participants are asked to complete the same assessment tasks both nition (ASR) cognition through dialogue-based content design.
before and after they are given intervention or are subjected to a con Dialogue-based content design involves acoustic, lexicon and language
dition. These tasks are used to determine whether or not there are any models (Lestari & Nugraha, 2017). After the ASR system was initially
changes in the participants’ performance that could be attributed to the trained, the lexicon and a corpus of audio data are leveraged as input to
treatment or condition (Ryu & Cheong, 2017). A survey study is a means generate the language and acoustic models. Audio content will be dis
of gathering information or data as reported by participants (Misje, played on the screen consisting of tangible human-like dialogue that
Bosnes, Gåsdal, & Heier, 2005). A questionnaire study is a research in performs natural interactions (Lestari & Nugraha, 2017).
strument that consists of a set of questions that is used to gather
meaningful information from respondents. The structure of these in
struments is similar to that of an interview, and it consists of questions 4.1. Technological integrations
that can be responded to either written or spoken (Misje et al., 2005). A
mixed-method is a research approach whereby quantitative and quali Technological integration refers to the use of technology for
tative data are collected within the context of the same study (Pluye, improving students’ learning experience (El Shazly, 2021). This princi
Gagnon, Griffiths, & Johnson-Lafleur, 2009). ple is particularly important in language learning, where technology has
Two major technical grounds were discovered among the dialogue been shown to increase student engagement and motivation, resulting in
system studies: technologies for establishing human–dialogue system improved learning outcomes. By effectively integrating AI technologies,
relationships and embedded cognition via AI content design. The first students can develop essential skills such as critical and creative
theoretical grounding was the interactive relationship between humans thinking, collaboration, communication, and problem-solving, which
and the AI dialogue system. An AI dialogue system is able to autono are essential for success in the 21st century. Therefore, incorporating
mously communicate with students from various educational back technology is crucial in enhancing student learning and success (Pham &
grounds (Moussalli & Cardoso, 2021, p. 226) by fostering students’ Sampson, 2022; Wu et al., 2022; Xia, 2020). The technology behind
readiness toward EFL interactional practices and also enhancing their chatbots includes AI, natural language processing (NLP), and machine
learning (El Shazly, 2021). There are various technologies involved in
Fig. 3. Task design of dialogue systems.
13
improving students’ pronunciation, aiming for successful spoken inter adopted an IPA dialogue system for the speaking and listening devel
action, e.g., ASR, audiolingual, IPA, SDS, CALL, DMS, CILLE, D3, ITS, opment of EFL English students. Findings show that students’ interac
and RAI & GAI. tional competence was considerably improved with IPA task design.
Several studies used more than one language training approach at Five studies included SDS in the dialogue system, and the findings
the same time in the design of interactional practice activities. The most showed that dialogue systems enhance students’ performance in
often used method was automatic speech recognition (ASR) (n 8), fol speaking. Two studies included CALL in the dialogue systems, and
lowed by audiolingual (n = 5), intelligent personal assistant (IPA) (n = findings showed that low-level EFL students benefit the most, and their
5), spoken dialogue systems (SDS) (n = 5), computer-assisted language speaking proficiency improved dramatically after interacting with the
learning (CALL) (n = 3), dialogue management systems (DMS) (n = 2), system. CALL reinforces and evaluates the content that has to be taught.
cognitive immersive language learning environment (CILLE) (n = 1), It often consists of a substantial number of interactive components.
data-driven development (D3) (n = 1), intelligent tutoring system (ITS) Two studies adopted DMS, and the findings showed that dialogue
and the retrieval-based AI & generative AI (RAI & GAI) (n = 1). Table 3 systems motivate EFL students’ willingness to communicate. One study
illustrates the technological integration embedded in the dialogue sys introduced a CILLE in which an AI dialogue system enables multi-modal
tem, which includes Automatic Speech Recognition, Audiolingual, In interactions to better immerse language learning. The dialogue system
tellectual Personal Assistant, Spoken Dialogue System, and Computer- in CILLE provides a sense of being unintrusive when supporting multi-
Assisted Language Learning. party interactions. One study has adopted D3 in the dialogue system,
ASR is a cutting-edge technology that enables machines to convert Enskill®English, to foster foreign language learning, and the findings
speech signals into text or commands after speech recognition. Audio suggested that Enskill®English is useful for acquiring spoken English
lingual incorporates a drill routine of listening and speaking in language abilities. One study included ITS in the dialogue system to trigger
acquisition, Via the audiolingual method, dialogue systems identify and constructive interaction among students, and the results found that the
adapt individual students’ interactive behaviors to successfully engage ‘weak’-directed agent intervention mode seems to foster a more
students in the target language use (Hanif, 2016). IPA is a software agent constructive discourse among students. One study, as illustrated in
that performs tasks or provides services for a user’s instructions or in Table 3, embedded RAI & GAI in the dialogue system to enhance learner
quiries. CALL is a method of teaching and learning in which content engagement in computer-supported collaborative learning, and findings
learned is presented, reinforced, and assessed using computer and indicated that the dialogue system improved student engagement in
computer-based resources. SDS is a computer system capable of collaboration tasks.
conversing with a user. It consists of two essential components: a voice Several studies revealed that language technological integration in
recognizer and a text-to-speech component. DMS are the NLP compo the design of interactional practices in dialogue systems for university
nents responsible for analysing and contextualising user-to-chatbot in education place a considerable emphasis on ASR (Chen et al., 2020;
teractions. D3 is a method of software development that uses data to Dizon, 2020; Najima et al., 2021) and audiolingual systems (Wu et al.,
direct the development process. ITS refers to computer learning envi 2020). Based on the literature, the ASR system provides students with
ronments that assist students in acquiring information and skills. ITS additional opportunities for creating output in the target language. Most
adopts powerful intelligent algorithms that adapt to the learner on a EFL students praised the diversity of task activities and the ease with
fine-grained level and execute complicated learning concepts. The which the ASR can be accessed. Students believed that the ASR inte
retrieval-based AI uses keyword matching to choose the most relevant grated system could foster their interactional competence and enhance
answer in both machine and deep learning, whereas generative based AI vocabulary memorization and retention. Moreover, the AI dialogue
uses predefined responses to generate new dialogue based on the system embedded with ASR provides an excellent learning atmosphere
conversational training data. for EFL students who feel self-conscious about speaking in public.
Eight studies have embedded ASR in the dialogue systems. There are Several studies showed that the ASR-based dialogues systems could
two main components of ASR for EFL acquisition: (1) the ability to provide a low-anxiety environment to foster students’ speaking. One of
distinguish accented from incorrect pronunciation and (2) the ability to their preferred speaking strategies is the innovative negotiation for
offer the relevant evaluation on the quality of pronunciation. These meaning strategies in EFL acquisition. Table 4 provides brief review of
studies show that the dialogue systems improved students’ pronuncia findings on the interactional practice dialogue system for university
tion and vocabulary, and students’ vocabulary retainment and conver education in alignment with research questions.
sation skills significantly improved.
Five studies have adopted audiolingual function, whereby AI dia 4.2. Task design
logue systems were designed to improve EFL students’ interaction and
oral communication. Findings show that dialogue systems were able to Task design for language acquisition is a method facilitating lan
identify and adapt individual students’ interactive behaviors to suc guage learning that emphasizes communication and the usefulness of
cessfully engage students in the target language use. Five studies have the language rather than explaining the language forms. Language stu
dents are encouraged to complete tasks that require to utilize the foreign
Table 3 language or second language (Hismanoglu & Hismanoglu, 2011). Task
Technological integration embedded in dialogue system. design is critical for language learning because well-designed tasks can
promote effective language learning outcomes (Yu & Chen, 2018). Chen,
Approach Authors
Bear, Hui, Santhi-Ponnusamy, and Meurers (2022) claim that language
Automatic Speech Chen et al. (2020), Divekar et al. (2021, pp. 1–29), learning tasks should be designed in a way that promotes meaningful
Recognition Dizon (2017), Dizon (2020), El Shazly (2021),
Mohammadzadeh and Sarkhosh (2018), Morino et al.
and authentic communication, engages learners, and provides oppor
(2017, pp. 367–373), Najima et al. (2021) tunities for practice and feedback. Effective AI task design should also be
Audiolingual Dizon (2017), El Shazly (2021), Kim (2017), Li and aligned with learning goals and objectives and take into account
Peng (2021), Wu et al. (2020) learners’ interests, backgrounds, and proficiency levels. In
Intellectual Personal Chen et al. (2020), Dizon and Tang (2019), Hsu et al.
digitally-mediated learning activities, AI task design is particularly
Assistant (2021), Moussalli and Cardoso (2021), Wu et al. (2020)
Spoken Dialogue System Divekar et al. (2021); Mohammadzadeh and Sarkhosh crucial in terms of goal setting, feedback, and balancing skill difficulty.
(2018), Morino et al. (2017, pp. 367–373), Najima By paying careful attention to task design, language teachers can ensure
et al. (2021), Timpe-Laughlin et al. (2022) that language learning tasks are engaging, relevant, and aligned with
Computer-Assisted Mazzilli (2021), Yin and Satar (2020), Zou et al. (2020) learning objectives, thus promoting effective language learning out
Language Learning
comes (Liang, Yang, Shan, & Kim, 2021; Zhang, Zhao, Ouyang, & Zhao,
14
Table 4 Ten studies entailed negotiation of meaning strategy in language

Positive cognitive communication skills, and learning objectives. learning, which is a process through which speakers arrive at a mutually
Type of Cognition Contributing Elements to EFL Students’ Cognitive understanding of what has been said. Divekar et al. (2021, pp. 1–29)
Progression in University Education presented an immersive cognitive language learning environment
Retention Interactions with the AI dialogue systems for EFL that are embedded with negotiation for meaning functions to engage students in
reinforced by multi-modal stimuli on certain vocabulary a non-dyadic multi-modal environment which makes the EFL learning
items non-intrusive. The findings revealed that students’ proficiency in
Understanding EFL interaction with the dialogue systems leads to increased speaking increased dramatically between pre-test and post-test. EFL
concentration
Social-norms Students have the opportunity to get a deeper cultural
students were also exposed to cultural elements, which enriched their
awareness, and their intercultural competence is also understanding of the target language. In the immersive dialogue system,
enhanced students observed and experienced unfamiliar social practices and
Application Rephrasing commands when the system did not understand norms, via negotiation for meaning techniques students have the op
their utterances
portunity to get a deeper cultural awareness, and their intercultural
Analysis Asking questions, repeating or generating words or phrases
verbally, and conversing in role-playing scenarios competence is also enhanced. Students indicated that cultural absorp
tion fostered their willingness to communicate.
Few studies included question and answer design in the dialogue
Communication Contributing Elements to EFL Students’ Communicative
skills Skills in University Education systems, Najima et al. (2021) adopted the question-and-answer method
Interactional practice Shadowing practicing oral skills and mimicking a natural to improve students’ speaking proficiency, and findings indicated that
conversation the dialogue system is effective for language learning and further
Pronunciation Emphasize lexical items and sentence patterns in
revealed that the students’ eye gaze behaviors changed between
conversations
Vocabulary usage Vocabulary memorization, and everyday dialogues
question-answering tasks. The eye gaze behaviors act as an indicator of
Story-retelling Enhancing students’ productive communication and the level of concentration of the students. Three studies integrated sto
creative skills rytelling in the dialogue system to provide a sense of familiarity and
Negotiation for Involving asking for clarification, rephrasing, and verifying trust. The storytelling method engages the listener with the story and
meaning
makes them more receptive to learning. Morino et al. (2017, pp.
367–373) investigated the effectiveness of digital storytelling on Japa
Emotional state Contributing Elements to EFL Students’ Emotional States nese university EFL students’ attitudes and awareness toward CALL, and
with AI Dialogue Systems
Enjoyment Enjoyed interacting with AI Dialogue systems for EFL as they
the findings showed that storytelling increased students’ awareness of
are non-judgmental CALL and improved their speaking proficiency.
Relaxed Feel less anxious in a relaxing environment Studies showed that the task designs were aligned with language
Motivated Highly motivated in learning English with the use of an AI learning methods, such as negotiation for meaning and storytelling
EFL dialogue system
techniques (Timpe-Laughlin et al., 2022; C. H. Wu et al., 2022). In the
Frustrated Technological issues and unnatural TTS sound
literature, it shows that task design such as negotiation for meaning
embedded in synchronous computer-mediated communication has been
2016). extensively studied (Wu et al., 2020), and two additional stages,
The general functionalities from the reviewed studies include con confirmation and reconfirmation, have been extended to the process of
versation (n = 13), negotiation for meaning (n = 10), question & an negotiation routine (Timpe-Laughlin et al., 2022). At the confirmation
swers (n = 5), and storytelling (n = 3). The negotiation for meaning stage, students confirm with their counterparts that the issue is attended
method is believed to improve comprehension skills in the language to or unattended through a reaction to response which signals the stu
acquisition process, during which interactiveness plays a vital role in dents’ acceptance or remaining unaccepted. At the reconfirmation stage,
producing comprehensible input for students, asking for clarification, it indicates the closure of the negotiation routine signaling students’
and providing an opportunity to restructure modified output (Dizon, understanding (Timpe-Laughlin et al., 2022).
2020). A conversation function in a dialogue system is able to record,
understand, and analyze vocal or text input and respond in a chosen 4.3. Students’ engagement
language from a user (Xie et al., 2021). The question-and-answer
method in dialogue systems is a style of content delivery, and it con Learner engagement is a measure of the quantity and quality of a
sists of a chatbot asking questions and the students answering those learner’s involvement in their learning (Wu et al., 2020). Student
questions (Najima et al., 2021). Storytelling in dialogue systems is an engagement is essential for effective and practical language learning
innovative method that employs computer technology to enhance lan experiences (Tao et al., 2023). One of the effective ways to enhance
guage learning, and the technique assists the development of language student engagement is by incorporating AI through personalized and
and digital abilities in students (Morino et al., 2017, pp. 367–373). adaptive learning systems that can cater to the individual student’s
Several studies showed that conversation practise helps language needs, interests, and proficiency levels. This view is supported by Diwan,
students to assimilate their learned information, and integrate a range of Srinivasa, Suri, Agarwal, and Ram (2023) who claimed that students are
cognitive abilities to produce meaningful spoken communication. more likely to participate in peer-to-peer learning, and collaboration
Mohammadzadeh and Sarkhosh (2018) integrated conversation func facilitated by AI. When students are fully engaged in their studies, stu
tion in an AI dialogue system to improve students’ speaking abilities, dents tend to seek out additional learning opportunities when they are
and the results showed that the dialogue system enhances students’ fully engaged in their studies, leading to better language learning out
performance in speaking. Wu et al. (2020) carried out a study on uni comes. The higher that the students are engaged in their learning, the
versity EFL students’ attitudes towards an IPA dialogue system in rela more motivated, interested, and committed they are towards their
tion to interactional practice. The findings revealed that EFL speakers studies. Furthermore, student engagement promotes a positive learning
placed a significant emphasis on adjustments intended to increase the environment where students feel supported, challenged, and valued
possibility of being intelligible. This may be due to students’ speech through the assistance of AI. This positive AI environment, in turn,
limitations in the target language. Although students enjoyed the fosters a sense of belonging and enhances learning outcomes (Diwan
interactional practices with a dialogue system, students felt frustrated as et al., 2023; Thomas, Sarma, Gajula, & Jayagopi, 2022; Yang, Chen,
they had to exert a considerable effort to rephrase commands when the Flanagan, & Ogata, 2022).
system did not understand their utterances. Several studies show that students are more inclined to participate in
15
well-orchestrated speaking practice using a contextualized and appli the most appropriate AI tools and resources to support their language
cable way to the dialogue (Divekar et al., 2021, pp. 1–29; Wu et al., learning needs, and avoid wasting time and resources on technologies
2020; Xie et al., 2021). In addition to facilitating bi-directional that may not be effective or appropriate for their language learning
communication (Xie et al., 2021) with verbal, gestural, and physical context (Ayedoun et al., 2019; Divekar et al., 2021, pp. 1–29).
interaction to enable students to practice receptive (such as under When students input incomplete phrases, failed communication was
standing words and language) (Timpe-Laughlin et al., 2022) and pro also discovered, or gibberish outputs were produced from dialogue
ductive (such as actively speaking and writing) language skills (Divekar systems (Ayedoun et al., 2019). Students’ positive emotional states (e.g.,
et al., 2021, pp. 1–29), language instructors were present to provide curiosity) in language acquisition tend to diminish with the absence of
procedural and learning support to students throughout the interactive emotion and visual indications during interactions (El Shazly, 2021).
tasks. In most cases, students participated in productive language ex Dialogue systems with weak AI were unable to interpret students’ in
ercises such as asking questions, repeating or generating words or puts. The dialogue system, for example, is unable to distinguish new
phrases verbally, and conversing in role-playing scenarios. subjects provided by students because their thoughts veer in unforeseen
Language instructors assisted them with a variety of tasks, such as ways as the interaction progresses (Divekar et al., 2021, pp. 1–29).
game introduction and feedback, and students were also able to learn Moreover, they are isolating language students from the language
from their experience to improve their learning outcomes (Bashori et al., learning environment due to dialogue systems’ odd robotic voices and
2020). Xie et al. (2021) incorporated an immersive learning environ incapacity to have lengthy conversations. Divekar et al. (2021, pp. 1–29)
ment called OpenSim to improve student engagement, and the system observed that students expressed frustration with technological glitches
also tracks individual students’ study curves. Bashori et al. (2020) and experienced exhaustion after absorbing the instructional content.
claimed that EFL students speaking anxiety was dramatically reduced
via interacting with the dialogue system, and their engagement was 4.6. The novelty effect
enhanced in the anxious-free environment. Timpe-Laughlin et al. (2022)
highlight that EFL students enjoyed the interaction with the dialogue A novelty effect occurs when the results of a research study are
system with speaking activities as they received immediate feedback. influenced by the novelty or newness of a particular intervention or
product being used. This effect can make it difficult to determine
4.4. Learning objectives whether the outcomes of the study are actually effective or just a result
of the novelty factor (Fryer et al., 2017). AI dialogue systems are not
Cognitive learning objectives from interactional practice with a immune to this effect, and it has been noted that learners may experi
dialogue system were associated with significant (a) academic ence short-term gains in engagement and performance due to the nov
achievement (Morino et al., 2017, pp. 367–373), such as understanding, elty effect. This novelty effect may temporally boost students’
application and analysis, (b) increased concentration (Najima et al., motivation or learning performance due to the technology novelty
2021), such as retention, (c) enhanced cultural awareness (Divekar (Fryer et al., 2017). According to Fryer et al. (2017), this effect occurs
et al., 2021, pp. 1–29) such as social-norms and 4) learning through with the dialogue system during a 16-week experimental study. Stu
collaboration (Divekar et al., 2021, pp. 1–29) such as association. dents’ interest in speaking tasks reported decreased slowly after the first
Learning objectives are a critical component of effective learning communication task. Similarly, Lubis, Sakti, Yoshino, & Nakamura
because they provide a clear description of what students are expected (2019) argued that the dialogue system was perceived as a curiosity by
to achieve as a result of their instruction. They guide the development of the students rather than as a long-term collaborator in their everyday
effective plans for instruction, activities, and assessments (Al Braiki, language practice, and delivering a workshop before the first session
Harous, Zaki, & Alnajjar, 2020). Incorporating AI in learning objectives may help lessen the novelty impact. All studies suggested that a pro
can help ensure that students acquire the necessary knowledge and skills longed period of the experiment will diminish the novelty effect, and in
to succeed in today’s fast-paced technological environment. With clear this way, students will benefit from the advantages of technological
learning objectives, students can better focus their efforts and achieve innovation of EFL acquisition. To minimize the impact of the novelty
their learning goals, leading to successful application of their newly effect in AI-based dialogue systems, teachers and researchers can
acquired knowledge and skills (Wong, Ma, Dillenbourg, & Huan, 2020). consider delivering a workshop or training session before using the
The communicative abilities of students, such as interactional practice, technology to familiarize learners with the AI and integrate it as a
pronunciation, vocabulary usage, story-retelling, and negotiation for long-term collaborator in language practice. This way, students can
meaning were significantly improved (Morino et al., 2017, pp. benefit from the advantages of AI innovation for EFL acquisition over a
367–373). In particular, the speed at which students speak and their prolonged period of time (Lubis et al., 2019).
response accuracy is enhanced (Wu et al., 2020), such as grammatical
(Manda, Nurlaila, & Indri, 2017), and lexical correctness (Bashori et al., 5. Discussion
2020). Table 4 highlights the impact of dialogue systems on students’
emotional state, feelings of enjoyment, relaxation, and motivation. Q1: The key evaluation dimensions of AI dialogue systems for
enhancing EFL in the university
4.5. Technological limitation
Six dimensions were identified in the study which includes (1)
Although a dialogue system can assist students in their language technological integration, (2) task designs, (3) students’ engagement,
acquisition, it is vital to recognize the limits of their technical compe (4) learning objectives, (5) technological limitation, and (6) the novelty
tence. The apparent unnaturalness of the computer-generated voice, effect, as shown in Table 2. The finding indicates that AI dialogue sys
which students compared to humans, was the most commonly reported tems serve a communicative purpose of EFL acquisition in university
technical obstacle (McTear et al., 2016). AI continues to advance in education as students find dialogue systems alleviate their speaking
language learning, it is crucial to understand its limitations to enable anxiety, enhance their vocabulary retainment and offer immediate
learners to make informed decisions about the use of specific technol feedback to boost their willingness to interact in the target language.
ogies for supporting their learning goals (Bresciani, Dabić, & Bertello, However, one paper finds that AI dialogue systems increase students’
2022, p. 102063). Zhai and Wibowo (2022) argue that understanding anxiety instead of reducing it.
technological limitations allows learners to avoid unrealistic expecta
tions and frustration, and instead focus on leveraging AI technology
effectively. By understanding the limitations of AI, learners can identify
16
5.1. Technological integrations linguistic dimensions of the actors’ lines (Alsadoon, 2021). EFL students
were willing to participate in interactive language learning activities
There are various technologies involved in improving students’ such as storytelling and role-play using communicative, meaning-based
pronunciation, aiming for successful spoken interaction, several studies language learning approaches (Divekar et al., 2021, pp. 1–29). Although
used more than one language training approach at the same time in the it is important to note that students’ spoken output tended to be
design of interactional practice activities, five studies have adopted closed-ended answers at both the lexical and syntax levels, this indicates
audiolingual function, whereby AI dialogue systems were designed to that future efforts should be directed toward developing tasks that
improve EFL students’ interaction and oral communication. Two studies emphasize intelligibility in order to fulfil meaning-focused instruction.
adopted DMS, and the findings showed that dialogue systems motivate In this systematic review, studies (Dizon & Tang, 2019; Morino et al.,
EFL students’ willingness to communicate. Several studies, as summa 2017, pp. 367–373) find that the integration of storytelling in AI dia
rized in Table 3, place a considerable emphasis on ASR to enhance logue systems is construed as fitting technologically with task-based,
students’ interactional competence. project-based, and thematic strategies, and this is because the story
However, not all studies alleviate EFL students’ speaking anxiety, telling technique is consistent with critical and dialogic approaches of
enhance their vocabulary retainment and offer immediate feedback to learning, in which space is made for collective thinking, experimenta
boost their willingness to interact in the target language. El Shazly tion, and peer review (Morino et al., 2017, pp. 367–373). In Morino
(2021) argues that foreign language anxiety has been a recurring issue in et al.’s (2017, pp. 367–373) study, authors expect students to craft a
the domain of EFL acquisition. In his study, the author investigated the narrative in a bilingual form because it prevents the compartmentali
role of the AI dialogue system embedded with ASR in speaking practice zation of languages and it acknowledges students’ interrelationships
from 48 Egyptian university students enrolled in an EFL course. Over between what is in mind and what is in usage. This makes it possible for
eight weeks of a quasi-experimental study, he found that students suf students’ bilingual or multilingual repertoires to be represented in a
fered dramatically from foreign language anxiety, notably in their holistic manner and for syncretic identities to be built. This helps
speaking capabilities which were at similar percentages between stimulate university EFL students’ confidence and encourages them to
pre-test and post-test measurements. This situation may be derived from place a higher value on their bilingual or multilingual skills. Yin and
concerns about failing the course instead of fostering students’ interests Satar (2020) found that the lack of non-linguistic elements, such as
in EFL acquisitions. EFL students’ affective status has been compromised paralinguistic gestures and facial expressions, imposes an undue burden
by the high-stake study under the academic context. Thus, students’ on written disclosure, delineating more creative clarity. This type of
final results may have inflicted severe constraints on their self-image, forced output in synchronous computer-mediated communication, ac
preventing students from engaging in communicative activities. El cording to Swain’s’s (1995) Comprehensible Output Hypothesis, forced
Shazly (2021) also argues that students’ anxiety levels were not allevi output increases students’ awareness of the linguistic characteristics of
ated when being exposed to a non-threatening AI dialogue system, and their interlanguage. When students encounter a gap in their linguistic
this finding contrasts with the literature that AI dialogue systems posi knowledge of the target language, learning occurs (Yin & Satar, 2020).
tively enhance students’ engagement with non-judgmental AI dialogue Based on the retrieved studies, task design, such as everyday con
systems (El Shazly, 2021). versation and storytelling, are still prevalently used in EFL learning. Yet,
On the other hand, technological integration of an audiolingual EFL students in university education are required to gain a body of in
method is used to develop bottom-up language skills, such as word formation from various subject domains and disciplines. Thus, problem-
recognition (Ayedoun et al., 2019). In word association exercises, stu solving skills are needed for students to cultivate self-direct learning
dents were able to gain receptive language skills, such as vocabulary and skills. Problem-solving skills encompass a wide range of constructivist
phrases. In addition, students participated in constructive language perspectives and stretch over many fields as university students are able
usage by administering story-retelling tasks. Such engagement enhanced to attend to issues from a real-world perspective. Students are able to
students’ productive communication (e.g., spoken communication construct new linguistic knowledge with a more profound comprehen
abilities) and creative skills (Morino et al., 2017, pp. 367–373). How sion and better retention outcomes (Azman & Shin, 2012).
ever, based on the studies reviewed, the audiolingual method, which is Thus, a well-designed EFL dialogue system for university education
also known as rote learning, is the most vital component in acquiring can have more linguistic content through various techniques such as
new vocabulary. A few studies (Mei, 2018; Saydaliyeva, Atamirzayeva, debate, discussion or problem-solving. For example, Google’s Dialog
& Dadaboyeva, 2020) found that repetition and imitation lead to limited flow endorses users to modify conversational content by alternating pre-
linguistic development in university education. As the audiolingual set datasets (e.g., intents) (Sagar et al., 2021). BotStar, an online dia
method is mechanical associative learning, it ignores constructivist logue system platform, enables users to create conversational flows by
language learning, which emphasizes collaboration and idea exchanges dragging and dropping intents on a designed dashboard (Shevat, 2017).
(Saydaliyeva et al., 2020). Studies find that university EFL students are These functions allow EFL instructors to script students’ learning expe
inclined to construct their own linguistic knowledge of the material riences in accordance with the desired learning goals (Sagar et al., 2021;
related to their own experiences. Learning is enhanced when university Shevat, 2017).
students are encouraged to establish views and develop their thoughts
about the linguistic materials being studied (El Shazly, 2021; Morino 5.3. Students’ engagement
et al., 2017, pp. 367–373; Saydaliyeva et al., 2020).
Learner engagement is a measure of the quantity and quality of a
5.2. Task design learner’s involvement in their learning (Wu et al., 2020). Multiple
studies demonstrate that students are more likely to engage in
Task design is a systematic way in which task description, workflow well-organized speaking practice that is contextualized and related to
and task execution plan are structured. Task design for language the dialogue system (Wu et al., 2020). In most cases, students partici
acquisition helps to facilitatee language learning by emphasizing on the pated in productive language exercises such as asking questions,
regular communication and the usefulness of the language rather than repeating or generating words or phrases verbally, and conversing in
explaining the language forms (Hismanoglu & Hismanoglu, 2011). role-playing scenarios.
Based on the literature, language acquisition can be achieved by However, no studies have introduced or mentioned with what kinds
adopting storytelling techniques for such purposes as (a) learning vo of technology were embedded to foster EFL students’ engagements so as
cabulary meaning related to the content of a story, (b) formulating to enhance to learning outcomes. One suggestion might be the integra
individualized questions related to the story, and (c) repeating certain tion of humor, empathy practice and culture awareness in the dialogue
17
system to encourage students in a culturally humorous yet empathetic 5.4. Learning objectives
environment. Palmer and Menard-Warwick (2012) argue that humor,
empathy practice and culture awareness in class plays a significant role Cognitive learning objectives from interactional practice with a
in increasing the levels of psychological and emotional wellness which dialogue system were associated with significant academic achievement
in return help students achieve higher levels of academic accomplish (Morino et al., 2017, pp. 367–373), and to achieve better scores for
ment. The English language has been identified as a means of intercul certain English tests seems to be an utterly important goal for most
tural communication between individuals of varied linguistic and university EFL students. English tests such as the International English
cultural origins (Zhang, Mandl, & Wang, 2011). Language Testing System (IELTS), Cambridge English Test, and Pear
Humor, the ability to recognize or communicate what is humorous, is son’s Test of English (PTE) Academic, have adopted communicative
both a source of amusement and a technique of dealing with tough or approaches with supervised interactions between the interviewer and
embarrassing circumstances and stressful occurrences (Thomas, Nelson, the interviewee or interactions between different interviewees (Mastura,
& Silverman, 2015). To have a sense of humor requires not just a 2021). Thus, interactional competence becomes an important compo
comprehension of the language and words, but also of how they are nent of the curriculum in EFL education because they allow students to
used, what they signify, the underlying culture, the connotations, and interact effectively with others in the target language (Timpe-Laughlin
the unwritten messages (Thomas et al., 2015). et al., 2020). By developing interactional competence in handling lin
Empathy is described as the capacity to identify and understand the guistic queries, language learners gained greater leverage to optimize
feelings and thoughts of other people, as well as the ability to project their academic achievement (Mastura, 2021).
one’s own thoughts and feelings onto those of another (Matsuhisa et al., Interactional competence help to increase concentration and col
2021). Empathy is the language that allows language learners to speak laborations (Najima et al., 2021), this is similar to Young (2011)’s
with a level of depth that fosters connection and enables them to statement that the author claimed that interactional competence is
comprehend the logic behind another person’s decisions, regardless of distinct from communicative competence in that it places a greater
how divergent they may be to their own (Matsuhisa et al., 2021). emphasis on interactions on the development of spoken English.
Previous research has shown that the use of humor in the classroom Conversational competence is very context-dependent and
helps enhance students’ interactional skills (Gonulal, 2021; Hismanoglu, co-constructed by speakers who participate in the conversation. As a
Ersan, & Turan, 2018), and improves students’ understanding and result, it requires a larger focus and concentration on the relationships
retention of learning material (Farnia & Mohammadi, 2021; Gonulal, that occur between interlocutors. The capacity to deploy interactional
2018). Pomerantz and Bell (2011) advocated for the inclusion of humor resources (turn-taking, repair, boundaries, speech actions, etc.) using
not simply as a method for boosting enjoyment, but also as an essential accessible language resources as required by the speaker or the hearer in
component of the process of acquiring practical interactional abilities. order to convey their communicative intents in real situational cir
The authors argue that when it comes to verbal information, such as cumstances is what we mean when we talk about interactional compe
words, sentences, and visual information, learner’s memories are tence (Galaczi & Taylor, 2018).
improved through the use of humor. As a result of factors such as greater Interactional competence also enhances cultural awareness (Divekar
attention and enhanced memory rehearsal, the humor effect can be et al., 2021, pp. 1–29), which is in par with Hymes’ (1972) statement,
ascribed to improved encoding and retrieval of funny information as the author argued that not only does competence refer to an individual’s
contrasted to—and often at the price of—non-humorous information. knowledge of the forms and structures of language, but it also extends to
The perception of humor engages both hemispheres of the brain. The how an individual uses language in actual social and cultural situations.
left hemisphere is the “logical brain,” which is responsible for language Hymes outlined four categories of information that people draw upon
competence, logical analysis, and detailed humor recognition, while the while interacting in various social and cultural settings. These categories
right hemisphere is the “creative brain,” which is responsible for include what is conceivable to accomplish with language, what is
imagining and comprehending humor (Palmer & Menard-Warwick, practicable to do with language, what is proper to do with language, and
2012). As humor is universal yet tinted by culture (Palmer & what is really done according to the social norm.
Menard-Warwick, 2012), in an egalitarian framework, students from
horizontal collectivist cultures emphasize sociability and interdepen 5.5. Technological limitation
dence with others and practice affiliative humor. Meanwhile, students
from vertical collectivist cultures comply to the hierarchical social order Although a dialogue system can assist students in their language
stratified by their in-groups and are willing to conduct altruistic actions acquisition, it is vital to recognize the limits of their technical compe
for their in-groups, they prefer to use self-defeating humor. This is in tence. The apparent unnaturalness of the computer-generated voice,
contrast to a low individualist society (collectivism), in which values which students compared to humans, was the most commonly reported
such as self-sacrifice, interdependence, and generosity towards others technical obstacle (McTear et al., 2016).
are emphasized (Palmer & Menard-Warwick, 2012). In addition, Mem There are three obstacles in using the system. First, the inability of an
bers in collectivist cultures tend to focus more on others than on AI dialogue system to interact with students of heavily accented speech,
themselves, resulting in a society with a far deeper sense of empathy and and thus students’ interactional competence may be compromised
care for others (Palmer & Menard-Warwick, 2012). (Chen et al., 2020). Interactional practice is arguably the most critical
Regarding empathy, it is the capacity to perceive the experiences of component of EFL development (Najima et al., 2021). Many language
others from various cultures, including both high context and low students may be unable to engage in AI dialogue interaction due to the
context societies (Matsuhisa et al., 2021). A person forms and defines his fact that some AI dialogue systems appear to have difficulty fully un
or her sense of self by conforming to certain society norms, values, and derstanding EFL speech with non-standard English variants. This high
beliefs. Students from high-context cultures are expected to discover the lights a broader problem with AI-assisted dialogue, and speech
hidden complexity from their culturally ingrained intuitions, whereby recognition accuracy across various language variants, take English di
they perceive low-context communication as overly detailed. In alects as an example, Tatman (2017) finds that YouTube automatic
dividuals who appreciate a low-context communication style, on the captioning systems struggle to perform its accuracy with Scottish En
other hand, see high-context conversationalists as lacking in trans glish. Johnson (2019) states that the choices of the dialogue system are
parency or incapable of effectively transmitting ideas (Matsuhisa et al., too limited and restricted, and encountered bugs, and the author
2021). believed that natural language understanding was designed to under
stand the simple language of students. Thus, suggestions proposed by
Divekar et al. (2021, pp. 1–29) were that language students should be
18
cognizant of technical restrictions, and this is because dialogue systems effectiveness of the technology (Fryer et al., 2017). Therefore, re
operate by using pre-determined subjects and pre-stored phrases, which searchers and educators must be cautious in interpreting the results of
affect natural language processing. Thus, students should try to avoid studies involving novel technologies and consider the potential impact
colloquialisms and partial sentences. of the novelty effect on the outcomes. The study by Lubis et al. (2019)
Second, unnaturalness of the computer-generated voice was reported highlights the potential negative impact of the novelty effect on the use
technical obstacle (McTear et al., 2016), which tends to refrain EFL of AI dialogue systems for language practice. Learners may perceive the
learners participation of interaction with the dialogue system. Usually, technology as a curiosity rather than a long-term collaborator, limiting
synthetic speech is often created from text, and due to a revolution in AI, their ability to fully engage with the technology and benefit from its
the audio quality of synthetic speech has significantly improved over the features (Fryer et al., 2017). Therefore, to minimize the impact of the
last several years. Elmers, Werner, Muhlack, Möbius, and Trouvain novelty effect on the outcomes of studies involving new technologies,
(2021) suggest that a breath noise is embedded into the system to reduce appropriate training and support should be provided to learners.
the unnaturalness of the computer-generated voice which showed that Engagement with AI dialogue systems can be sustained by making it
there is an improvement in students’ recollection for sentences that were meaningful and useful for students, and overcoming the novelty effect
preceded by a breath noise as opposed to sentences that did not include a that can cause engagement to fall once the initial excitement wears off.
breath noise. The authors also found that a significant influence for Tsay, Kofinas, Trivedi, and Yang (2020) suggests that gamified learning
phrase length, arguing that shorter sentences had more accuracy for systems with gamified elements can attract users initially, but engage
recollection than longer ones when breath noises were included. ment can fall once the novelty effect wears off. Longitudinal iterative
Third, dialogue systems with weak AI were unable to interpret stu cycles are important for observing the novelty effect and generating
dents’ incomplete inputs, when students’ ideas wander unpredictably as ways to overcome its potential negative impact on engagement. The
the engagement proceeds (Divekar et al., 2021, pp. 1–29). The inability study found that for a substantial proportion of students, engagement
to interpret students inputs also isolate language learners from the with the gamified learning system was intrinsically driven, which helped
language-learning environment as a result of dialogue systems cannot to overcome the novelty effect. This resulted in engagement becoming
extend further discussions. Divekar et al. (2021, pp. 1–29) report that habitual, playful, and game-like, and helped to sustain engagement over
students suffered weariness after digesting instructional information and time.
showed irritation with the dialogue system cannot interpret students’
inputs. Suggestions to increase the response accuracy from incomplete 5.7. Summary of the six dimensions
input can follow most popular classification approach to use imputation
to replace incomplete values with plausible values (Elmers et al., 2021). Most studies indicated that technological integration in dialogue
Tran, Zhang, Andreae, Xue, & Bui (2018) believe that in many systems for EFL improves students’ interactional competence for both
real-world datasets, it is impossible to prevent incomplete inputs or within and outside of the classroom. As task design in the dialogue
missing values, and authors provided novel methods for merging system systematically delivers task description, and organises workflow
imputation, clustering, and feature selection for classification with and task execution plan by emphasizing on frequent communication and
partial data in order to enhance efficiency without sacrificing precision. the utility of the language rather than teaching the language forms to aid
Clustering is used to decrease the number of examples utilized by the language learning. Then, using learner engagement as a tool to quantify
imputation. Feature selection is used to eliminate duplicate and un the extent and quality of a learner’s participation in their EFL, as well as
necessary characteristics of training data, which significantly decreases to elucidate the knowledge and skills that students have attained from a
the cost of imputation. The research also studies Differential Evolution’s certain course or programme. As the development and implementation
(DE) capacity to find feature subsets with inadequate data. The results of an AI dialogue system in EFL is still in its infancy stage, technological
demonstrate that integrating imputation, clustering, and feature selec limitations occur. These limitations include the inability of computer
tion not only improves classification accuracy, but also significantly software or hardware to perform a certain function, as well as the
decreases the computation time necessary to estimate missing values novelty effect, which is when learners’ performance initially improves
when classifying new instances. when new technology is implemented as a result of increased interest in
Another recommendation to increase the response accuracy from the new technology.
incomplete inputs is to remove stop words and linking synonyms by This study has examined 28 articles on the interactive design of
using Natural Language Toolkit (NLTK), and the phrase “stop word” dialogue systems for EFL for university education. They include tech
refers to frequent phrases that do not add to any deeper meaning and nological integration, task design, students’ engagement, and learning
that the dialogue systems are trained to disregard (Kaur & Buttar, 2018). objectives. Based on the study, there is overwhelming support for En
Wordnet’s definition of synonyms is that they are “words that convey glish language learning in university education. Students acquire
the same notion and are interchangeable in various situations.” As a declarative and procedural knowledge through constant repetition or
result of this, synonyms are categorized into sets that are not arranged in rehearsal of linguistic activities, and they get personalized support in a
any particular order (synsets). These synsets are what we make use of in mistake-friendly atmosphere via a dialogue system, which enhances
order to generate the synonyms and antonyms that are shown in the students’ language abilities without time and space constraints (Divekar
following programs. When stop words are removed, the low-level in et al., 2021, pp. 1–29). When students get an opportunity to apply their
formation is eliminated, which allows the dialogue systems focus on key linguistic skills verbally in a simulated real-life environment, it reduces
features of the output. Eliminating stop words does not only reduce the their anxiety and enhances their desire to converse in the target lan
vector space, but also improves performance overall. guage (Timpe-Laughlin et al., 2020). Nonetheless, this type of language
learning engagement seems to be diminishing significantly over time as
5.6. The novelty effect students’ interest fade away due to a novelty effect, the lower value of
dialogue systems compared to human facilitators (Bashori et al., 2020).
The novelty effect is a phenomenon that can occur in studies However, challenges still remain in relation to offering students tangible
involving new or innovative technologies, such as dialogue systems, assistance to become expressive and eloquent in speaking discourse
where the initial excitement and curiosity generated by the technology (Divekar et al., 2021, pp. 1–29).
can lead to enhanced engagement and performance among learners From the literature, AI dialogue systems in EFL learning in university
(Hammad & Bahja, 2023; Miguel-Alonso, Rodriguez-Garcia, Checa, & education are used to offer synchronous formative feedback to students
Bustillo, 2023). However, it is important to note that the gains made by about their spoken performance and to stimulate their metacognitive
learners may be short-term, and the novelty effect can mask the true reasoning strategies by analyzing students’ learning patterns (Dizon,
19
2020). The studies found that the majority of students are from countries This is because of the numerous cognitive qualities that young people
where English is a foreign language, and countries such as China, Japan, have. These cognitive traits have an effect on how an EFL student takes
and South Korea embrace a vertical collectivist culture that conforms to in information, how that knowledge is perceived, and how it is orga
the hierarchical social system stratified by their in-groups and are pre nized in their minds.
pared to perform altruistic acts for their in-group. This contrast with The use of culturally relevant humor in EFL class is proven to
students from horizontal collectivist cultures emphasizes sociability and improve the long-term outcomes and wellbeing. Chabeli (2008) argues
interdependence with others in an egalitarian context (Zhang et al., that cross-cultural humor enhanced the emotional intelligence and
2011). critical thinking of learners. In an egalitarian framework, students from
Regardless of whether learning takes place within and outside of the horizontal collectivist cultures emphasize sociability and interdepen
classroom, EFL students can get access to essential information through dence with others and practice affiliative humor. Whereas students from
dialogue systems. Forsythe (2017) referenced what Young (2011) stated vertical collectivist cultures comply to the hierarchical social order
in the situated learning theory that learning takes place in various places stratified by their in-groups and are willing to conduct altruistic actions
and times. Students have access to an abundance of language support for their in-groups, they prefer to use self-defeating humor. This is in
materials and authentic situations or activities via mobile phones and contrast to a low individualist society (collectivism), in which values
the Internet. On the other hand, An, Gan, and Wang (2020) argue that such as self-sacrifice, interdependence, and generosity towards others
students may regulate their own pace through dynamic and spontaneous are emphasized (Palmer & Menard-Warwick, 2012). Critical thinking is
self-regulated learning based on cognitive, emotional, motivational, and a sort of cognitive capacity that is essential for relatively rapid and
behavioral components. In An et al.‘s (2020) self-regulated learning practical learning, and it entails the development of flexibility, which
theory, the authors believed that self-generated ideas, emotions, and aids in the analysis and evaluation of knowledge after it has been ac
behaviors are geared towards the achievement of a learner’s own ob quired. Learners who have acquired good critical thinking skills are
jectives. Such elements enable students to adapt their behaviors to meet often in a continual state of inquiry, striving to establish logical expla
their learning goals in a variety of educational contexts. Consequently, nations of their experience and to negotiate their prior knowledge
students may manage and adapt their knowledge and plans to be suit (Thomas, 2011). It is advised that incorporating culturally relevant
able with the ever-changing contexts or settings for learning. In foreign comedy models is beneficial for establishing a dynamic learning envi
language acquisition, An et al. (2020) state that language students may ronment that enables students to make fair judgements and build critical
achieve more success in their language learning and usage if they are thinking abilities.
more strategic in their efforts. To achieve an effective dialogue system for EFL, this paper recom
mends that it is vital to include cultural considerations in the behavioral
Q2: Gaps arising from the results, additional necessary study, and models of dialogue systems. The content and the form of dialogue sys
recommendations to achieve effective systems tems’ utterances aimed toward certain cultures must be consistent. For
instance, depending on the culture simulated by the dialogue system,
To increase EFL students’ long-term learning outcomes, instead of various dimensions for a small chat could be chosen, and different
short-term knowledge acquisition for just proficiency test, the integra discourse markers can be deployed to express politeness. Based on the
tion of humor, empathy practice and culture awareness in class plays a diverse cultural backgrounds, the volume and quality of motions may
significant role in increasing the levels of psychological and emotional also vary. The perception and selection of communication behaviors are
wellness which in return help students achieve higher levels of academic heavily influenced by often overlooked cultural factors (Endrass &
accomplishment and interactional competence (Palmer & André, 2014).
Menard-Warwick, 2012). It has been discovered that 28 studies seemed If cultural dimensions are not taken into consideration while devel
overlooked the cultural backgrounds, emotional recognition, and humor oping interactive behaviors for a dialogue system, this may cause an
aspects in the design of dialogue system for EFL aimed for enhancing adverse effect on user acceptability. Several studies, such as Shiban et al.
university students’ interactional competence. The study also found that (2015), showed on how virtual agents were perceived racial distinctions
the importance of debate and problem-solving skills seemed was not in their verbal and nonverbal communication. Researchers altered the
considered in the design of AI dialogue system, thus additional necessary behavior of the virtual agent to mimic the behavior of people of various
studies are needed in order to achieve effective dialogue systems for EFL. races, but the agent’s appearance remained racially undefined
throughout the project. Researchers analyzed students’ views of the
5.8. Lack of cultural considerations virtual agent’s ethnicity by having them complete a questionnaire.
Students were able to appropriately classify the virtual agent to various
Despite the fact that the learner’s cultural background plays a role in ethnicities based on its behavioral manners. They tried to reflect their
the formation of the EFL learner’s identity as well as the cognitive ethnicity in a way that can be used for educational purposes. Even if
characteristics a learner adopts, none of 28 studies have considered the culture is not explicitly considered during the modelling process of a
importance of learners’ cultural backgrounds in designing the dialogue virtual agent, the character will nevertheless exhibit characteristics of a
system for EFL. According to Benfilali, Nadif, Khartite, Benattabou, and particular cultural background. This is usually that of the programmer
Bouih (2021), acquiring a second language is a way for a person to since they are the one who determines whether or not the agent’s
exhibit their culture as well as their cultural identity. A method to EFL behavior is natural. When a programmer comes from a Western culture
learning incorporates intercultural language learning, which encom where direct eye contact is thought to reflect honesty, this may seem to
passes helping students acquire an awareness of their own culture as be unacceptable to a person from a different cultural background, such
well as an identification with other cultures. In terms of cognitive as a member of Asian culture, and this behavior may be considered
characteristics, Mehrotra and Yilmaz (2015) contend that the cognitive unfriendly by others.
characteristics coexist in the development of language proficiency. The
first cognitive characteristic is an analytical procedural system that is 5.9. Lack of empathetic considerations
rule-based, and the second cognitive characteristic is a declarative sys
tem that is example-based. Adults are more likely to be defined by the Most studies found that dialogue systems were able to identify and
former, whilst children are defined by the latter when learning English adapt individual students’ interactive behaviors to engage students in
as a foreign language. Young people naturally have a greater potential the target language use. However in EL Shazly’s (2021) study, the
for learning than their adult counterparts do, therefore they have an author investigated the role of the AI dialogue system embedded with
advantage when it comes to the process of natural language acquisition. ASR in speaking practice from 48 Egyptian university students enrolled
20
in an EFL course. Over eight weeks of a quasi-experimental study, he Palmer and Menard-Warwick (2012) argue that humor is universal yet
found that students suffered dramatically from foreign language anxiety, tinted by culture. The capacity of the brain as a whole may be increased
notably in their speaking capabilities which were at similar percentages through the use of humor. Similarly McGhee (1983) argues that the
between pre-test and post-test measurements. Under such circumstance, perception of humor stimulates both hemispheres of the brain at the
no solution, maybe empathetic encouragement, was provided in EL same time. The left hemisphere is known as the “logical brain,” and it is
Shazly’s study to alleviate anxiety or stress when EFL learners experi responsible for language proficiency, logical analysis, and detailed
ence a significant negative impact during their speaking training session humor recognition. On the other hand, the right hemisphere is known as
while using a dialogue system. According to Gennaro et al.’s (2020) the “creative brain,” and it is responsible for imagining and under
findings, dialogue systems that include empathic responses have a standing humor. Learners who participated in activities that required
palliative impact on individuals who are nervous. There have been convergent thinking showed a tendency to engage both hemispheres of
various studies using empathic dialogue systems to elicit appropriate the brain while working on a creation task, and learners who partici
emotional responses to the emotional condition of students learning pated in activities that required humor exhibited brain activity that was
English in settings other than universities (Fryer, Coniam, Carpenter, & comparable to that of learners who engaged in convergent thinking
Lăpușneanu, 2020; Santos, Ethel, & Resurreccion, 2020). The authors activities (Thomas et al., 2015).
believed that it is essential to include culturally related empathic To achieve an effective dialogue system for EFL, it is recommended
modelling into dialogue systems for EFL so that cognitive and affective that future research devote considerable effort to finding the impact of
components can capture and react to university EFL students’ emotional cross-cultural humor on the language ability of EFL students. Thomas
state regardless of where they are learning (Marx & Pray, 2011). et al. (2015) find that increased in learners’ performance and good re
Thus, incorporating empathy modelling in dialogue systems for EFL sults were observed when humor approaches were utilized in the
would be helpful for resolving issues caused by learners from two classroom. The research found that one of the most advantages of
distinct cultural backgrounds and would aid EFL learners in adjusting to incorporating cross-cultural humor into the classroom is the improve
new environments. According to Cundiff, Nadler, and Swan (2009), ment of learning and knowledge recall abilities. Baker, D’Mello,
empathy encompasses cultural values, ideas, and perceptions, similar to Rodrigo, and Graesser (2010) claimed that the continuity of learning,
the culturally related humor. Benattabou (2020) establishes the notion which includes cognitive and emotional states, happens in a particular
of the unity of theory and reality after examining the link between the context. The authors suggested that the cognitive and emotive states of a
concept of empathy and pragmatics. He then subdivided empathy into learner are crucial when culture-related humor is used effectively during
pragma-linguistic value and socio-pragmatic value within the field of learning. Learners are likely to boost their self-motivation and be
intercultural communication. The author came to the conclusion that it actively engaged in their studies if the learning material and goal are
is necessary to combat ethnocentrism, negative stereotypes, and preju entertaining and engaging (Hoa, 2011).
dice, as well as acknowledge the equality of different cultural traditions. Based on the study carried out by El Shazly (2021), students suffered
Therefore, including culturally appropriate empathic modelling into dramatically from foreign language anxiety, notably in their speaking
conversation systems for EFL will provide students a better under capabilities which were at similar percentages between pre-test and
standing of cultures other than their own and erase psychological bar post-test measurements. To encourage and motivate weak EFL students,
riers that exist between different languages and cultural norms. we recommend that intercultural empathy functions are needed as
To achieve an effective dialogue system for EFL to further enhance cultural empathy intends to improve a relationship or to understand
learners’ learning outcomes, and the design of the dialogue systems for communication effectively, empathy contains shared emotional re
EFL needs to include functions or algorithms to facilitate learners to sponses (affective) and perspective-taking (cognitive) activities. When
comprehend the experiences of individuals from other cultures, which the context is insinuated, it enables one to step into the shoes of others
includes both high context and low context civilizations (Matsuhisa based on the constituents of their internal frame of reference or world
et al., 2021). By adhering to specific societal standards, values, and view, such as beliefs, motivations, fears, and aspirations (Krznaric,
beliefs, an individual creates and defines his or her sense of self. Students 2010).
reared in high-context cultures are expected to discern the concealed The use of AI algorithms in cultural elements raises several ethical
complexity from their culturally established intuitions, and they observe and legal challenges. It requires lawmakers and media platforms to take
low-context communication as being excessively descriptive. In steps to regulate the transmission of cultural and creative content to
contrast, individuals who value a low-context style of communication prevent discrimination or isolation. Data used as inputs to the algo
characterize high-context conversationalists as lacking in transparency rithms include racial, gender, and stereotypes, which can have serious
or incapable of properly conveying ideas (Matsuhisa et al., 2021). repercussions. Ethics frameworks have recently been proposed to
encourage accountable, ethical, and accommodating algorithms of all
5.10. Lack of humor considerations stakeholders. Metcalf and Crawford (2016) argue that ethical frame
works must establish effective regulations as various social platforms so
As technological integration in this study focuses on primarily far cannot self-regulate.
improving students’ pronunciation; task design focuses on encouraging
students to complete tasks, such as negotiation for meaning, question & 5.11. The importance of debate and problem-solving
answers, and storytelling; students’ engagement focuses on learners’
involvement in their learning, no studies have focuses on dialogue for Through discussion or debate, the EFL students acquire greater un
EFL that improve students’ attitude toward the dialogue systems by derstanding through shared viewpoints and ideas (Mei, 2018). However,
developing culturally relevant humor, boosting their motivation and some technological integrations, such as audiolingual method cannot
enhancing their performance. In addition, there has been no research provide the effectiveness of discussion as it is a mechanical associative
conducted to characterize the nature of the humor mechanisms that are learning, as most studies in this systematic review primarily focused on
used by computer interfaces. This would involve, first, reflecting the interactional practices such as basic vocabulary memorization and
cultural background of the target language, and then, providing everyday dialogues. It is believed that debate is a potentially useful
contextual recognition strategies that guide conversations when they are educational technique for EFL speaking tasks since it can scaffold the
required. It is essential to include culturally relevant comedy since learning process in university education. During a debate stage, EFL
research on language instruction in the classroom suggests that doing so students take up the majority of attention and are required to
might aid boost students’ emotional intelligence. As a result, it is impromptu elaborate their ideas and viewpoints over the ‘clash’ period
essential to incorporate culturally relevant humor (Chabeli, 2008). of the debate (Mei, 2018). Debates improve students’ interactional
21
competence, in addition, their language, vocabulary, delivery, and identify students who need additional support and provide them
diction will improve so as to present their arguments concisely. Debates with personalized guidance.
help students to listen attentively and explain their ideas clearly (Mei, • Use the AI dialogue system to record and track various learning
2018). characteristics of each learner comprehensively. This will enable
teachers to gain insight into each student’s prior knowledge, specific
6. Conclusion learning style, preferences, and needs, and create personalized
teaching and learning strategies. This approach will result in a more
This paper presented an overview of the technological applications effective and engaging learning experience for each student.
of dialogue systems for EFL in university education and explored EFL • Incorporate the AI dialogue system to explore and pursue their
students’ interactional practices among various agents for EFL acquisi passions, leading to increased engagement and motivation. Teachers
tion. A systematic review of existing studies on AI dialogue systems was can utilize the AI dialogue system to incorporate students’ interests
conducted systematically between 2013 and 2022. In particular, the into the lessons, providing relevant hands-on activities, and offering
study questions focus on technological integrations, task designs, stu related learning resources with multimedia-rich content such as
dents’ engagement, and learning objectives. The results of this system videos, simulations, and games. This combination will help students
atic review indicate that the instructional design of dialogue systems better understand complex subjects and retain information more
makes use of communicative language learning, such as audiolingual efficiently.
approaches and narrative, as the most prominent language learning
techniques, which led to a positive skill-based cognitive learning 6.2. Limitations and future research possibilities
outcome.
Existing information and analyses are limited to interactional prac
tice for EF acquisition, which is one of the limitations of this research. In
6.1. Theoretical and practical implications
addition, the investigation is limited to material published between
January 2012 and August 2022. Future research should investigate the
This study has both theoretical and practical contributions. From
use of alternative technologies to enhance the interactional practices of
theoretical perspective, this systematic review can serve as a foundation
other businesses, such as the medical and hospitality sectors.
for learning new languages and other technological domains. Following
There is also a paucity of research regarding the cultural back
concerns from stakeholders, the current teaching of spoken English and
grounds of students while providing personalized feedback. In general,
interactional competence is still viewed as unsatisfactory. This is due to
the significance of variety in feedback while considering users’ cultural
the fact that many graduates still lack the ability to communicate in a
background is neglected. Feedback contexts and cultures are inter
foreign language, which prevents them from participating fully in
connected, and research on student feedback cultures is mostly influ
transcultural and global society (Beshir & Yigzaw, 2022; Darmajanti,
enced by cultural variance and shows students’ difficulties with
2017). The study has provided a profound understanding of the obsta
learning, critical thinking, and academic writing. Thirdly, research
cles and advantages arising from the interactional practices process with
demonstrates that students suffer dissatisfaction and stress throughout
dialogue systems in EFL interactional competence. Through the use of
the learning process (Ayedoun et al., 2019; Divekar et al., 2021, pp.
the dialogue systems, this review has presented an opportunity to
1–29), but no sympathetic approach is offered to ease these issues. In
perform more research that would aid in the improvement of the
overall, the preceding research demonstrate that dialogue systems have
learning experience of students in general. Recommendation to improve
a major favorable effect on the speaking instruction of students.
response accuracy from incomplete inputs can follow the most common
According to the findings of the study, several studies have been
categorization strategy and utilize imputation to replace missing values
conducted on the technical usage of dialogue systems for EF acquisition.
with reasonable values (Tran et al., 2018), and using NLTK to eliminate
Future study should enhance the learning experiences of students with
stop words and connect synonyms can further enhance the accuracy of
respect to the other three macro abilities, namely listening, reading, and
responses to partial inputs. When stop words are omitted, low-level in
writing. This study’s focus is limited to an investigation of conversation
formation is discarded, allowing conversation systems to concentrate on
systems in EFL acquisition. Age and gender should be the focus of future
crucial output characteristics. Not only does eliminating stop words
study, and ethnography will assist in the creation of successful learning
lower the vector space, but it also increases overall performance.
strategies.
Practically, this research sheds light on the obstacles of interactional
None of the papers included in this study was conducted longitudi
practice in employing a dialogue system in EFL acquisition. Prior to
nally. There was a wide range of intervention periods in the experiments
designing a dialogue system to maximize students’ outcomes, it is crit
reviewed, ranging from a few minutes to a few weeks, which might have
ical to consider spoken discourse methods, instructional design, the roles
resulted in a novelty effect in the language learning process. Studies
of dialogue systems, and skill-based cognitive learning techniques. The
spanning two or more academic years should be investigated in order to
study’s findings can be utilized by educators to build effective strategies
determine whether students’ interest or motivation in engaging with
for dialogue systems. The results are also useful for language instructors
dialogue systems increases or remains the same.
to better identify the situation associated with using new technology in
Finally, the majority of the existing studies on dialogue systems for
order to provide suitable ways for enhancing students’ learning expe
language learning relied on self-reported questionnaires for their data
riences and outcomes.
collection, which did not adequately illustrate the potential outcomes of
A few relevant teaching suggestions for on-site teaching teachers are
utilizing dialogue systems on students’ language success.
also presented below:
• Incorporate the AI dialogue system into their daily teaching routine. 6.3. Recommendations
This system can be used to assist students in finding information,
providing instant feedback and guidance, and answering routine This paper has examined the use of dialogue systems in interactional
questions on the class discussion forum. This approach can free up practice for EFL acquisition for university education. It is found that
teachers’ time, allowing them to concentrate on fostering students’ interactional practice for university education has not yet reached the
critical thinking and problem-solving skills, while students can stage of mainstream use of dialogue systems in language learning. The
analyze problems and access necessary information efficiently. By following suggestions are made for designing an EFL dialogue system to
using the AI dialogue system consistently over time, teachers can improve student’s experience:
22
⁃ A prolonged period of an experiment will diminish the novelty effect, Canale, M., & Swain, M. (1980). Theoretical bases of communicative approaches to
second language teaching and testing. Applied Linguistics, 1(1), 1–47.
and in this way, students will benefit from the advantages of tech
Chabeli, M. (2008). Humor: A pedagogical tool to promote learning. Curationis, 31(3),
nological innovation in EFL acquisition. 51–59.
⁃ Further studies are required to develop speaking activities for EFL Chen, X., Bear, E., Hui, B., Santhi-Ponnusamy, H., & Meurers, D. (2022). Education
students which are accessible and user-friendly at educational in theories and AI affordances: Design and implementation of an intelligent computer
assisted language learning system. In Artificial intelligence in education. Posters and
stitutions in university education for academic purposes. late breaking results, workshops and tutorials, industry and innovation tracks,
⁃ To guarantee that speaking activities are more effective, it is neces practitioners’ and doctoral consortium: 23rd international conference, AIED 2022 (pp.
sary to build a framework for deploying dialogue systems for uni 582–585). https://doi.org/10.1007/978-3-031-11647-6_120. Durham, UK, July
27–31, 2022, Proceedings, Part II.
versity education. Chen, H. H.-J., Yang, C. T.-Y., & Lai, K. K.-W. (2020). Investigating college EFL learners’
perceptions toward the use of Google Assistant for foreign language learning.
Interactive Learning Environments, 1–16. https://doi.org/10.1080/
Declaration of competing interest 10494820.2020.1833043
Chomsky, N. (1965). Aspects of the theory of syntax, 1965. Cambridge, MA: MIT Press.
The authors have no relevant financial or non-financial interests to Colquhoun, H. L., Levac, D., O’Brien, K. K., Straus, S., Tricco, A. C., Perrier, L., et al.
(2014). Scoping reviews: Time for clarity in definition, methods, and reporting.
disclose. Journal of Clinical Epidemiology, 67(12), 1291–1294.
Cundiff, N. L., Nadler, J. T., & Swan, A. (2009). The influence of cultural empathy and
References gender on perceptions of diversity programs. Journal of Leadership & Organizational
Studies, 16(1), 97–110.
Da Silva, J., Fernandes, V., Limont, M., & Rauen, W. B. (2020). Sustainable development
Adenle, Y. A., Chan, E. H., Sun, Y., & Chau, C. (2020). Exploring the coverage of
assessment from a capitals perspective: Analytical structure and indicator selection
environmental-dimension indicators in existing campus sustainability appraisal
criteria. Journal of Environmental Management, 260, Article 110147. https://doi.org/
tools. Environmental and Sustainability Indicators, 8, Article 100057. https://doi.org/
10.1016/j.jenvman.2020.110147
10.1016/j.indic.2020.100057
Darmajanti, P. (2017). The development of interactional competence of EFL learners. In
Al Braiki, B., Harous, S., Zaki, N., & Alnajjar, F. (2020). Artificial intelligence in
International conference on English language teaching (ICONELT 2017) (pp. 180–183).
education and assessment methods. Bulletin of Electrical Engineering and Informatics, 9
https://doi.org/10.2991/iconelt-17.2018.40
(5), 1998–2007. https://doi.org/10.11591/eei.v9i5.1984
Diederich, S., Brendel, A. B., Morana, S., & Kolbe, L. (2022). On the design of and
Al-Oqaily, E. I., Salam, A. R. H., & Na, K. S. (2022). The use of blackboard in the practice
interaction with conversational agents: An organizing and assessing review of
of English-Speaking skills among Saudi EFL learners during COVID-19. Arab World
human-computer interaction research. Journal of the Association for Information
English Journal. https://doi.org/10.24093/awej/covid2.23
Systems, 23(1), 96–138. https://doi.org/10.17705/1jais.00724
Alsadoon, R. (2021). Chatting with AI bot: Vocabulary learning assistant for Saudi EFL
Divekar, R. R., Drozdal, J., Chabot, S., Zhou, Y., Su, H., Chen, Y., et al. (2021). Foreign
learners. English Language Teaching, 14(6), 135–157. https://doi.org/10.5539/elt.
language acquisition via artificial intelligence and extended reality: Design and evaluation.
v14n6p135
Computer Assisted Language Learning. https://doi.org/10.1080/
An, Z., Gan, Z., & Wang, C. (2020). Profiling Chinese EFL students’ technology-based self-
09588221.2021.1879162
regulated English learning strategies. PLoS One, 15(10), Article e0240094. https://
Diwan, C., Srinivasa, S., Suri, G., Agarwal, S., & Ram, P. (2023). AI-based learning
doi.org/10.1371/journal.pone.0240094
content generation and learning pathway augmentation to increase learner
Arung, F., Rafli, Z., & Dewanti, R. (2019). In E. Setiawati, et al. (Eds.), A new substantive
engagement. Computers & Education: Artificial Intelligence, 4, Article 100110. https://
theory for the mastery of foreign language speaking skills: A grounded research (p. 231).
doi.org/10.1016/j.caeai.2022.100110
Ayedoun, E., Hayashi, Y., & Seta, K. (2019). Adding communicative and affective
Dizon, G. (2017). Using intelligent personal assistants for second language learning: A
strategies to an embodied conversational agent to enhance second language learners’
case study of alexa. TESOL Journal, 8(4), 811–830. https://doi.org/10.1002/tesj.353
willingness to communicate. International Journal of Artificial Intelligence in Education,
Dizon, G. (2020). Evaluating intelligent personal assistants for L2 listening and speaking
29(1), 29–57. https://doi.org/10.1007/s40593-018-0171-6
development. Language, Learning and Technology, 24(1), 16–26.
Ayedoun, E., Hayashi, Y., & Seta, K. (2020). Toward personalized scaffolding and fading
Dizon, G., & Tang, D. (2019). A pilot study of Alexa for autonomous second language
of motivational support in L2 learner–dialogue agent interactions: An exploratory
learning. CALL and Complexity, 107.
study. IEEE Transactions on learning technologies, 13(3), 604–616. https://doi.org/
El Shazly, R. (2021). Effects of artificial intelligence on English speaking anxiety and
10.1109/TLT.2020.2989776
speaking performance: A case study. Expert Systems, 38(3), Article e12667.
Azman, N., & Shin, L. K. (2012). Problem-based learning in English for a second language
Elmers, M., Werner, R., Muhlack, B., Möbius, B., & Trouvain, J. (2021). Take a breath:
classroom: Students’ perspectives. International Journal of Learning, 18(6). https://
Respiratory sounds improve recollection in synthetic speech. Interspeech,
doi.org/10.18848/1447-9494/CGP/v18i06/47648
3196–3200. https://doi.org/10.21437/Interspeech.2021-1496
Baker, R. S., D’Mello, S. K., Rodrigo, M. M. T., & Graesser, A. C. (2010). Better to be
Elo, S., & Kyngäs, H. (2008). The qualitative content analysis process. Journal of
frustrated than bored: The incidence, persistence, and impact of learners’
Advanced Nursing, 62(1), 107–115. https://doi.org/10.1111/j.1365-
cognitive–affective states during interactions with three different computer-based
2648.2007.04569.x
learning environments. International Journal of Human-Computer Studies, 68(4),
Endrass, B., & André, E. (2014). Integration of cultural factors into the behavioral models
223–241.
of virtual characters. Natural Language Generation in Interactive Systems. https://doi.
Bashori, M., van Hout, R., Strik, H., & Cucchiarini, C. (2020). Web-based language
org/10.1017/CBO9780511844492.010
learning and speaking anxiety. Computer Assisted Language Learning, 1–32. https://
Farnia, M., & Mohammadi, S. (2021). Exploring EFL teachers’ and learners’ perception of
doi.org/10.1080/09588221.2020.1770293
L2 humor: A case study of Iranian English language institutes. Eurasian Journal of
Benattabou, D. (2020). Helping EFL students avoid socio-pragmatic failure: Focus on
Applied Linguistics, 7(1), 151–168. https://doi.org/10.7592/EJHR2018.6.1.gonulal
nonverbal intercultural competence. TESOL and Technology Studies, 1(1), 23–41.
Firmstone, J. (2019). Editorial journalism and newspapers’ editorial opinions. In Oxford
Bener, E. (2021). Flipping EFL classes in higher education: A systematic review. Language
research encyclopedia of communication. https://doi.org/10.1093/acrefore/
Education and Technology, 1(2), 90–109, 0000-0002-8587-8538.
9780190228613.013.803
Benfilali, I., Nadif, B., Khartite, B., Benattabou, D., & Bouih, A. (2021). Cross gender oral
Forsythe, E. M. (2017). A qualitative case study of Japanese university students and personal
communication from biological difference and socialized identity to mutual
smartphone use in English as a Foreign Language classes. Northcentral University.
understanding. Journal of World Englishes and Educational Practices, 3(5), 13–27.
Fryer, L. K., Ainley, M., Thompson, A., Gibson, A., & Sherlock, Z. (2017). Stimulating and
https://doi.org/10.32996/jweep.2021.3.5.2
sustaining interest in a language course: An experimental comparison of Chatbot and
Beshir, M., & Yigzaw, A. (2022). Students’ self-repair in EFL classroom interactions:
Human task partners. Computers in Human Behavior, 75, 461–468.
Implications for classroom dynamics. Asian-Pacific Journal of Second and Foreign
Fryer, L., Coniam, D., Carpenter, R., & Lăpușneanu, D. (2020). Bots for language learning
Language Education, 7(1), 1–15. https://doi.org/10.1186/s40862-022-00153-6
now: Current and future directions. Language, Learning and Technology, 24(2), 8–22.
Bibauw, S., François, T., & Desmet, P. (2019). Discussing with a computer to practice a
Galaczi, E., & Taylor, L. (2018). Interactional competence: Conceptualisations,
foreign language: Research synthesis and conceptual framework of dialogue-based
operationalisations, and outstanding questions. Language Assessment Quarterly, 15(3),
CALL. Computer Assisted Language Learning, 32(8), 827–877. https://doi.org/
219–236. https://doi.org/10.1080/15434303.2018.1453816
10.1080/09588221.2018.1535508
Gennaro, M., Krumhuber, E. G., & Lucas, G. (2020). Effectiveness of an empathic chatbot
Böhm, A. (2004). 5.13 theoretical coding: Text analysis in grounded theory. A Companion
in combating adverse effects of social exclusion on mood. Frontiers in Psychology, 10,
to.
3061.
Bowers, B., & Schatzman, L. (2016). Dimensional analysis. In Developing grounded theory
Girgin, U., & Brandt, A. (2020). Creating space for learning through ‘Mm hm’in a L2
(pp. 86–126). Routledge.
classroom: Implications for L2 classroom interactional competence. Classroom
Bresciani, S., Dabić, M., & Bertello, A. (2022). Collaborative technological development for
Discourse, 11(1), 61–79. https://doi.org/10.1080/19463014.2019.1603115
addressing grand challenges: Opportunities, limitations, and new frameworks. Elsevier.
Gonulal, T. (2018). Investigating the potential of humour in EFL classrooms. The
Butler, Y. G. (2019). Assessment of young English learners in instructional settings. https://
European Journal of Humour Research, 6(1), 141–161. https://doi.org/10.7592/
doi.org/10.1007/978-3-030-02899-2_24. Second handbook of English language
EJHR2018.6.1.gonulal
teaching.
Gonulal, T. (2021). Investigating EFL learners’ humorous interactions with an intelligent
Camiciottoli, B. C., & Campoy-Cubillo, M. C. (2018). In Introduction: The nexus of
personal assistant. Interactive Learning Environments, 1–14. https://doi.org/10.1080/
multimodality, multimodal literacy, and English language teaching in research and
10494820.2021.1974489
practice in higher education settings (Vol. 77, pp. 1–9). Elsevier.
23
Gorjian, B. (2017). The evaluation of using computer assisted language learning (CALL) Kramsch, C. (1986). From language proficiency to interactional competence. The Modern
facilities in developing EFL among Islamic Azad University practitioners: The case of Language Journal, 70(4), 366–372.
computer literacy. Iranian Journal of Applied Language Studies (Vol. 9,, 35–50. Krznaric, R. (2010). Empathy and climate change: Proposals for a revolution of human
Proceedings of the first international conference on language focus. relationships. Future ethics: Climate Change and Apocalyptic Imagination, 153–172.
Gunnarsdóttir, I., Davidsdottir, B., Worrell, E., & Sigurgeirsdóttir, S. (2020). Review of Kuhail, M. A., Alturki, N., Alramlawi, S., & Alhejori, K. (2023). Interacting with
indicators for sustainable energy development. Renewable and Sustainable Energy educational chatbots: A systematic review. Education and Information Technologies,
Reviews, 133, Article 110294. https://doi.org/10.1016/j.rser.2020.110294 28(1), 973–1018. https://doi.org/10.1016/j.caeai.2021.100033
Guzey, S. S., Yildiz, E. A., Demir, M. C., & Aksu-Dunya, B. (2022). Impact of EDpuzzle use Lestari, D. P., & Nugraha, R. R. (2017). A spoken-based question answering system for
on the assessment and measurement course achievement. HAYEF: Journal of train route service using the frame-based approach. The Sixth International Conference
Education, 19(1), 52–61. on Electrical Engineering and Informatics, 1–6. https://doi.org/10.1109/
Hall, J. K., Hellermann, J., & Doehler, S. P. (2011). L2 interactional competence and ICEEI.2017.8312384
development. Multilingual Matters. https://doi.org/10.21832/9781847694072 Liang, F., Yang, Y., Shan, Z., & Kim, B.-S. (2021). Application of artificial intelligence
Hammad, R., & Bahja, M. (2023). Opportunities and challenges in educational chatbots. technology in leisure sports course. In 2021 2nd international conference on computers,
Trends, applications, and challenges of chatbot technology, 119–136. https://doi.org/ information processing and advanced education (pp. 772–775). https://doi.org/
10.4018/978-1-6684-6234-8.ch005 10.1145/3456887.3457065
Hanif, H. (2016). Using Rosetta Stone software as media in teaching English vocabulary Lianto, F. (2019). Grounded theory methodology in architectural research. Journal of
(an experimental study at SDN No. 02 Lhoksukon). Getsempena English Education Physics: Conference Series, 1179(1). https://doi.org/10.1088/1742-6596/1179/1/0
Journal, 2(1), Article 217667. https://doi.org/10.46244/geej.v2i1.681 12102.
Harati, H., Nooshinfard, F., Isfandyari-Moghaddam, A., Babalhavaeji, F., & Hariri, N. Li, K.-C., Chang, M., & Wu, K.-H. (2020). Developing a task-based dialogue system for
(2019). Factors affecting the unplanned use behavior of academic libraries users: English language learning. Education Sciences, 10(11), 306. https://doi.org/10.3390/
Towards an axial coding pattern. Aslib Journal of Information Management, 71(2), educsci10110306
138–154. https://doi.org/10.1108/AJIM-04-2018-0092 Li, B., & Peng, M. (2021). The evaluation of a blended teaching mode based on an AI language
Heift, T., & Schulze, M. (2015). Research timeline. https://doi.org/10.1017/ learning platform (pp. 1017–1021). The Second International Conference on
S0261444815000245 Information Science and Education. https://doi.org/10.1109/ICISE-
Hismanoglu, M., Ersan, Y., & Turan, Y. Z. (2018). Turkish EFL learners’ perceptions on IE53922.2021.00320
teachers’ using humor in the EFL classroom. Online Submission, 6(2), 284–294. Lubis, N., Sakti, S., Yoshino, K., & Nakamura, S. (2019). Positive emotion elicitation in
https://doi.org/10.18298/ijlet.2820 chat-based dialogue systems. IEEE/ACM Transactions on Audio, Speech, and Language
Hismanoglu, M., & Hismanoglu, S. (2011). Task-based language teaching: What every Processing, 27(4), 866–877. https://doi.org/10.1109/TASLP.2019.2900910.
EFL teacher should do. Procedia-Social and Behavioral Sciences, 15, 46–52. https:// Manda, R., Nurlaila, H., & Indri, W. (2017). Development of web-based computer-
doi.org/10.1177/003368820303400105 assisted language learning in English intensive course. IOP Conference Series:
Hoa, N. T. M. (2011). Developing EFL learners’ intercultural communicative competence: A Materials Science and Engineering, 180(1), Article 012291. https://doi.org/10.1088/
gap to be filled (Vol. 86). From defining EIL competence to designing EIL learning. 1757-899X/180/1/012291
Holton, J. A. (2007). The coding process and its challenges. The Sage handbook of Marx, S., & Pray, L. (2011). Living and learning in Mexico: Developing empathy for
grounded theory, 3, 265–289. English language learners through study abroad. Race, Ethnicity and Education, 14(4),
Hsu, Chen, H. H.-J., & Todd, A. G. (2021). Investigating the impact of the Amazon Alexa 507–535.
on the development of L2 listening and speaking skills. Interactive Learning Mastura, T. (2021). Assessing learners ‘communicative competence in different testing
Environments, 390–410. system. Online Journal of Sustainability and Leadership Research, 1(5), 154–158.
Hsu, Chen, & Yu, C.-S. (2021). Proposing a task-oriented chatbot system for EFL learners https://doi.org/10.1371/journal.pone.0233613
speaking practice. Interactive Learning Environments, 1–12. https://doi.org/10.1080/ Matsuhisa, T., Takahashi, N., Takahashi, K., Yoshikawa, Y., Aomatsu, M., Sato, J., et al.
10494820.2021.1960864 (2021). Effect of physician attire on patient perceptions of empathy in Japan: A
Huang, W., Hew, K. F., & Fryer, L. K. (2022). Chatbots for language learning—are they quasi-randomized controlled trial in primary care. BMC Family Practice, 22(1), 1–9.
really useful? A systematic review of chatbot-supported language learning. Journal of Mazzilli, F. (2021). Using Elbot to enhance conflict-solving skills in learners of German as
Computer Assisted Learning, 38(1), 237–257. https://doi.org/10.1111/jcal.12610 a foreign language. Educazione Linguistica. Language Education, 10. https://doi.org/
Huo, J. (2022). The role of learners’ psychological well-being and academic engagement 10.30687/ELLE/2280-6792/2021/01/005
on their grit. Frontiers in Psychology, 504. https://doi.org/10.3389/ McGhee, P. E. (1983). The role of arousal and hemispheric lateralization in humor. In
fpsyg.2022.848325 Handbook of humor research (pp. 13–37). Springer.
Huth, T., & Betz, E. (2019). Testing interactional competence in second language McTear, M., Callejas, Z., & Griol, D. (2016). Conversational interfaces: Past and present.
classrooms: Goals, formats and caveats. In Teaching and testing L2 interactional In The conversational interface (pp. 51–72). Springer. https://doi.org/10.1007/978-3-
competence (pp. 322–356). Routledge. https://doi.org/10.4324/9781315177021-13. 319-32967-3_4.
Hymes, D. (1972). On communicative competence, 269293J. B. Pride, & J. Holmes Mehrotra, R., & Yilmaz, E. (2015). Terms, topics & tasks: Enhanced user modelling for
(Eds.). Sociolinguistics, 269–293. better personalization. Proceedings of the 2015 International Conference on the Theory
Inkster, B., Sarda, S., & Subramanian, V. (2018). An empathy-driven, conversational of Information Retrieval, 131–140. https://doi.org/10.1145/2808194.2809467
artificial intelligence agent (Wysa) for digital mental well-being: Real-world data Mei, Y. (2018). Comments on the audiolingual method. International Journal of Arts and
evaluation mixed-methods study. JMIR mHealth and uHealth, 6(11), Article e12106. Commerce, 7(4), 90. https://doi.org/10.1007/978-3-319-32967-3_4
https://doi.org/10.2196/12106 Metcalf, J., & Crawford, K. (2016). Where are human subjects in big data research? The
Insights, T. B. (2021). English Language Learning market size by end user (corporate learners, emerging ethics divide. Big Data & Society, 3(1), 2053951716650211 https://doi.
individual users, educational institutions, & government institutions), methodology (online org/10.1177/2053951716650211.
learning, offline/classroom learning, & blended learning), global industry analysis, share, Miguel-Alonso, I., Rodriguez-Garcia, B., Checa, D., & Bustillo, A. (2023). Countering the
growth, trends, and forecast 2022 to 2030. https://www.thebrainyinsights.com/report novelty effect: A tutorial for immersive virtual reality learning environments. Applied
/english-language-learning-market-12711. Sciences, 13(1), 593. https://doi.org/10.3390/app13010593
Jafari, N., & Ansari, D. N. (2012). The effect of collaboration on Iranian EFL learners’ Misje, A. H., Bosnes, V., Gåsdal, O., & Heier, H. E. (2005). Motivation, recruitment and
writing accuracy. International Education Studies, 5(2), 125–131. retention of voluntary non-remunerated blood donors: A survey-based questionnaire
Johnson, W. L. (2019). Data-driven development and evaluation of Enskill English. study. Vox Sanguinis, 89(4), 236–244.
International Journal of Artificial Intelligence in Education, 29(3), 425–457, 0.1007/ Moghaddam, A. (2006). Coding issues in grounded theory. Issues in Educational Research,
s40593-019-00182-2. 16(1), 52–66.
Kampittayakul, T. (2019). The role of translanguaging in improving Thai learners’ Mohammadzadeh, A., & Sarkhosh, M. (2018). The effects of self-regulatory learning
interactional competence in dyadic" English as a foreign language" tutorial sessions. through computer-assisted intelligent tutoring system on the improvement of EFL
PASAA: A Journal of Language Teaching and Learning in Thailand, 56, 80–111. learners’ speaking ability. International Journal of Instruction, 11(2), 167–184.
Kaur, J., & Buttar, P. K. (2018). A systematic review on stopword removal algorithms. Morino, A., Lopez, E., & Ono, Y. (2017). Effects of a digital storytelling project on Japanese
International Journal on Future Revolution in Computer Science & Communication EFL Learners’ CALL attitudes and awareness of CALL tasks. Society for Information
Engineering, 4(4), 207–210. Technology & Teacher Education International Conference. https://www.learntech
Kawinkoonlasate, P. (2020). Online Language learning for Thai EFL learners: An analysis lib.org/primary/p/177309/.
of effective alternative learning methods in response to the covid-19 outbreak. Morse, J. M., Bowers, B. J., Charmaz, K., Corbin, J., Clarke, A. E., & Stern, P. N. (2016).
English Language Teaching, 13(12), 15–26. https://doi.org/10.5539/elt.v13n12p15 Developing grounded theory: The second generation (Vol. 3). Routledge. https://doi.
Kim, N.-Y. (2016). Effects of voice chat on EFL learners’ speaking ability according to org/10.1111/j.1548-1387.2010.01143.x
proficiency levels. Multimedia-Assisted Language Learning, 19(4), 63–88. https://doi. Moscardini, A., Strachan, R., & Vlasova, T. (2022). The role of universities in modern
org/10.15702/mall.2016.19.4.63 society. Studies in Higher Education, 47(4), 812–830. https://doi.org/10.1080/
Kim, N.-Y. (2017). Effects of types of voice-based chat on EFL students’ negotiation of 03075079.2020.1807493
meaning according to proficiency levels. English teaching, 72(1), 159–181. Moussalli, S., & Cardoso, W. (2021). Intelligent personal assistants and L2 pronunciation
Kim, H.-S., Cha, Y., & Kim, N. Y. (2021). Effects of AI chatbots on EFL students’ development: Focus on English past-. CALL and Professionalisation.
communication skills. Korean Journal of English Language and Linguistics, 21, Najima, T., Kato, T., Tamura, A., & Yamamoto, S. (2021). Remote learning of speaking in
712–734. https://doi.org/10.15738/kjell.21..202108.712 syntactic forms with robot-avatar-assisted language learning system. The 24th
Kools, S., McCarthy, M., Durham, R., & Robrecht, L. (1996). Dimensional analysis: International Conference on Text, Speech, and Dialogue, 558–566. https://doi.org/
Broadening the conception of grounded theory. Qualitative Health Research, 6(3), 10.1007/978-3-030-83527-9_48
312–330. Nirwana, N., & Suhono, S. (2022). Phonological interference in English pronunciation
produced by students at senior high school (A case study of Buginese and Javanese
24
students). Anglophile Journal, 3(1), 1–13. https://doi.org/10.51278/anglophile. Timpe-Laughlin, V., Sydorenko, T., & Daurio, P. (2020). Using spoken dialogue
v3i1.475 technology for L2 speaking practice: What do teachers think? Computer Assisted
Palmer, D. K., & Menard-Warwick, J. (2012). Short-term study abroad for Texas Language Learning, 1–24. https://doi.org/10.1080/09588221.2020.1774904
preservice teachers: On the road from empathy to critical awareness. Multicultural Timpe-Laughlin, V., Sydorenko, T., & Dombi, J. (2022). Human versus machine:
Education, 19(3), 17–26. Investigating L2 learner output in face-to-face versus fully automated role-plays.
Petersen, K. A. (2010). Implicit corrective feedback in computer-guided interaction: Does Computer Assisted Language Learning, 1–30. https://doi.org/10.1080/
mode matter? Georgetown University, 10822/553155. 09588221.2022.2032184
Pham, S. T., & Sampson, P. M. (2022). The development of artificial intelligence in Tran, C. T., Zhang, M., Andreae, P., Xue, B., & Bui, L. T. (2018). Improving performance
education: A review in context. Journal of Computer Assisted Learning, 38(5), of classification on incomplete data using feature selection and clustering. Applied
1408–1421. https://doi.org/10.1111/jcal.12687 Soft Computing, 73, 848–861. https://doi.org/10.1016/j.asoc.2018.09.026.
Pluye, P., Gagnon, M.-P., Griffiths, F., & Johnson-Lafleur, J. (2009). A scoring system for Tricco, A. C., Lillie, E., Zarin, W., O’Brien, K. K., Colquhoun, H., Levac, D., et al. (2018).
appraising mixed methods research, and concomitantly appraising qualitative, PRISMA extension for scoping reviews (PRISMA-ScR): Checklist and explanation.
quantitative and mixed methods primary studies in mixed studies reviews. Annals of Internal Medicine, 169(7), 467–473.
International Journal of Nursing Studies, 46(4), 529–546. Tripathi, M., Ray, N., Sinha, A., & Ahmed, F. (2021). New media and language learning:
Pomerantz, A., & Bell, N. D. (2011). Humor as safe house in the foreign language Opening global doors to linguistic development. International Journal of English
classroom. The Modern Language Journal, 95, 148–161. https://doi.org/10.1111/ Learning & Teaching Skills, 3(4), 2507–2566. https://doi.org/10.15864/Ijelts.3407
j.1540-4781.2011.01274.x Tsay, C. H. H., Kofinas, A. K., Trivedi, S. K., & Yang, Y. (2020). Overcoming the novelty
Roever, C., & Kasper, G. (2018). Speaking in turns and sequences: Interactional effect in online gamified learning systems: An empirical evaluation of student
competence as a target construct in testing speaking. Language Testing, 35(3), engagement and performance. Journal of Computer Assisted Learning, 36(2), 128–146.
331–355. Ulum, Ö. G. (2020). A critical deconstruction of computer-based test application in
Rosé, C. P., & Ferschke, O. (2016). Technology support for discussion based learning: Turkish State University. Education and Information Technologies, 25(6), 4883–4896.
From computer supported collaborative learning to the future of massive open online Underwood, J. H. (1984). Linguistics, computers, and the language teacher. A communicative
courses. International Journal of Artificial Intelligence in Education, 26(2), 660–678. approach. ERIC.
https://doi.org/10.1007/s40593-016-0107-y Vollstedt, M., & Rezat, S. (2019a). An introduction to grounded theory with a special
Ryu, E., & Cheong, J. (2017). Comparing indirect effects in different groups in single- focus on axial coding and the coding paradigm. Compendium for Early Career
group and multi-group structural equation models. Frontiers in Psychology, 8, 747, Researchers in Mathematics Education, 13(1), 81–100. https://doi.org/10.1007/978-
2017-25858-001. 3-030-15636-7
Sagar, R. H., Ashraf, T., Sharma, A., Goud, K. S. R., Sahana, S., & Sagar, A. K. (2021). Vollstedt, M., & Rezat, S. (2019b). An introduction to grounded theory with a special
Revolution of AI-enabled health care chat-bot system for patient aistance (Vol. 778). focus on axial coding and the coding paradigm. Compendium for Early Career
Springer. https://doi.org/10.1007/978-981-16-3067-5_18 Researchers in Mathematics Education, 13, 81–100. https://doi.org/10.1007/978-3-
Santos, O., Ethel, & Resurreccion, R. (2020). Therapist vibe: children’s expressions of 030-15636-7_4
their emotions through storytelling with a chatbot. Proceedings of the Interaction Wahyuningsih, S., & Afandi, M. (2020). Investigating English speaking problems:
Design and Children Conference, 483–494. https://doi.org/10.1145/ Implications for speaking curriculum development in Indonesia. European Journal of
3392063.3394405 Educational Research, 9(3), 967–977. https://doi.org/10.12973/eujer.9.3.967
Saydaliyeva, M., Atamirzayeva, E., & Dadaboyeva, F. (2020). Modern methods of Wang, Y., Feng, X., Guo, J., Gong, S., Wu, Y., & Wang, J. (2022). Benefits of affective
teaching English in Namangan state university. International Journal on Integrated pedagogical agents in multimedia instruction. Frontiers in Psychology, 12, 6639.
Education, 3(1), 8–9. https://doi.org/10.31149/ijie.v3i1.256 https://doi.org/10.3389/fpsyg.2021.797236
Shevat, A. (2017). Designing bots: Creating conversational experiences (Vol. 1). O’Reilly Weizenbaum, J. (1976). Computer power and human reason: From judgment to calculation.
Media, Inc. Wolfswinkel, J. F., Furtmueller, E., & Wilderom, C. P. (2013). Using grounded theory as a
Shiban, Y., Schelhorn, I., Jobst, V., Hörnlein, A., Puppe, F., Pauli, P., et al. (2015). The method for rigorously reviewing literature. European Journal of Information Systems,
appearance effect: Influences of virtual agent features on performance and 22(1), 45–55. https://doi.org/10.1057/ejis.2011.51
motivation. Computers in Human Behavior, 49, 5–11. Wollny, S., Schneider, J., Di Mitri, D., Weidlich, J., Rittberger, M., & Drachsler, H.
Skogmyr Marian, K., & Balaman, U. (2018). Second language interactional competence (2021). Are we there yet?-A systematic literature review on chatbots in education.
and its development: An overview of conversation analytic research on interactional Frontiers in artificial intelligence, 4.
change over time. Language and Linguistics Compass, 12(8), Article e12285. https:// Wong, G. K., Ma, X., Dillenbourg, P., & Huan, J. (2020). Broadening artificial intelligence
doi.org/10.1111/lnc3.12285 education in K-12: Where to start? ACM Inroads, 11(1), 20–29. https://doi.org/
Siddaway, AP, Wood, AM, & Hedges, LV (2019). How to do a systematic review: A best 10.1145/3381884
practice guide for conducting and reporting narrative reviews, meta-analyses, and Wright, M. (2022). Building connections in 2021–22. https://www.britishcouncil.org/
meta-syntheses. Annual Review of Psychology, 70, 747–770. https://doi.org/10.1146/ about-us/how-we-work/review-2021-22.
annurev-psych-010418-102803 Wu, C. H., Lin, H.-C. K., Wang, T.-H., Huang, T.-H., & Huang, Y.-M. (2022). Affective
Skogmyr, K., & Balaman, U. (2018). Second language interactional competence and its mobile language tutoring system for supporting language learning. Frontiers in
development: An overview of conversation analytic research on interactional change Psychology, 13, Article 833327.
over time. Language and Linguistics Compass, 12(8), Article e12285. https://doi.org/ Wu, Y., Rough, D., Bleakley, A., Edwards, J., Cooney, O., Doyle, P. R., et al. (2020). See
10.1111/lnc3.12285 what I’m saying? Comparing intelligent personal assistant use for native and non-
Swain’s, M. (1995). Three functions of output in second language learning. Principles and native language speakers. In Proceedings of the 22nd international conference on
Practice in Applied Linguistics: Studies in Honor of HG Widdowson, 125–144. human-computer interaction with mobile devices and services (pp. 1–9). https://doi.org/
Tai, T.-Y. (2022). Effects of intelligent personal assistants on EFL learners’ oral 10.1145/3379503.3403563
proficiency outside the classroom. Computer Assisted Language Learning, 1–30. Wu, Shirkey, G., Celik, I., Shao, C., & Chen, J. (2022). A review on the adoption of AI, BC,
https://doi.org/10.1080/09588221.2022.2075013 and IoT in sustainability research. Sustainability, 14(13), 7851. https://doi.org/
Tao, X., Shannon-Honson, A., Delaney, P., Dann, C., Xie, H., Li, Y., et al. (2023). Towards 10.3390/su14137851
an understanding of the engagement and emotional behaviour of MOOC students Xia, P. (2020). Application scenario of artificial intelligence technology in higher education
using sentiment and semantic features. Computers & Education: Artificial Intelligence. , international conference on applications and techniques in cyber intelligence ATCI
Article 100116. https://doi.org/10.1016/j.caeai.2022.100116 2019. In Applications and techniques in cyber intelligence (Vol. 7).
Tatman, R. (2017). Gender and dialect bias in YouTube’s automatic captions. Proceedings Xie, Q. (2017). Investigating the target language usage in and outside business English
of the First ACL Workshop on Ethics in Natural Language Processing, 53–59. https://doi. classrooms for non-English major undergraduates at a Chinese university. Cogent
org/10.18653/v1/W17-1606 Education, 4(1), Article 1415629. https://doi.org/10.1080/
Tegos, S., Demetriadis, S., & Tsiatsos, T. (2014). A configurable conversational agent to 2331186X.2017.1415629
trigger students’ productive dialogue: A pilot study in the CALL domain. International Xie, T., Liu, R., Chen, Y., & Liu, G. (2021). Moca: A motivational online conversational
Journal of Artificial Intelligence in Education, 24(1), 62–91. https://doi.org/10.1007/ agent for improving student engagement in collaborative learning. IEEE Transactions
s40593-013-0007-3S on learning technologies, 14(5), 653–664.
Thomas. (2011). Developing first year students’ critical thinking skills. Asian Social Yang, A. C., Chen, I. Y., Flanagan, B., & Ogata, H. (2022). How students’ self-assessment
Science, 26–35. behavior affects their online learning performance. Computers & Education: Artificial
Thomas, B., & Skinner, H. (2012). Dissertation to journal article: A systematic approach. Intelligence, 3, Article 100058. https://doi.org/10.1016/j.caeai.2022.100058
Education Research International, 2012, 1–11 (I) https://doi.org/10.1155/2012/ Yin, Q., & Satar, M. (2020). English as a foreign language learner interactions with
862135. chatbots: Negotiation for meaning. International Online Journal of Education and
Thomas, Nelson, J. K., & Silverman, S. J. (2015). Research methods in physical activity (7 Teaching, 7(2), 390–410. https://iojet.org/index.php/IOJET/article/view/707.
ed.). USA: Human kinetics Publisher. Young, R. F. (2011). Interactional competence in language learning, teaching, and
Thomas, C., Sarma, K. P., Gajula, S. S., & Jayagopi, D. B. (2022). Automatic prediction of testing. In Handbook of research in second language teaching and learning (pp.
presentation style and student engagement from videos. Computers & Education: 426–443). Routledge.
Artificial Intelligence, 3, Article 100079. https://doi.org/10.1016/j. Yu, Y., & Chen, Y. (2018). Design and development of high school artificial intelligence
caeai.2022.100079 textbook based on computational thinking. Open Access Library Journal, 5(9), 1–15.
Timpe-Laughlin, V., & Dombi, J. (2020). Exploring L2 learners’ request behavior in a https://doi.org/10.4236/oalib.1104898
multi-turn conversation with a fully automated agent. Intercultural Pragmatics, 17(2), Zhai, C., & Wibowo, S. (2022). A systematic review on cross-culture, humor and empathy
221–257. https://doi.org/10.1515/ip-2020-0010 dimensions in conversational chatbots: The case of second language acquisition.
Heliyon. , Article e12056. https://doi.org/10.1016/j.heliyon.2022.e12056
25
Zhai, C., Wibowo, S., & Cowling, M. (2022). Work-in-Progress—embedding cross-cultural Zhang, Y., Zhao, M., Ouyang, X., & Zhao, Y. (2016). Study on knowledge engineering
humorous and empathetic functions to facilitate language acquisition. 2022 8th education technique. In The 2016 international conference on education (pp. 27–31). E-
International Conference of the Immersive Learning Research Network (iLRN). https:// learning and Management Technology. https://doi.org/10.2991/iceemt-16.2016.6.
doi.org/10.1016/j.heliyon.2022.e12056 Zou, B., Liviero, S., Hao, M., & Wei, C. (2020). Artificial intelligence technology for EAP
Zhang, J., & Jing, Y. (2022). Application of artificial intelligence technology in cross- speaking skills: Student perceptions of opportunities and challenges. In Technology
cultural communication of intangible cultural Heritage. Mathematical Problems in and the psychology of second language learners and users (pp. 433–463). Springer.
Engineering. https://doi.org/10.1155/2022/6563114, 2022. Zrekat, Y., & Al-Sohbani, Y. (2022). Arab EFL University learners’ perceptions of the
Zhang, J., Mandl, H., & Wang, E. (2011). The effect of vertical–horizontal factors hindering them to speak English fluently. Journal of Language and Linguistic
individualism–collectivism on acculturation and the moderating role of gender. Studies, 18(1), 775–790. https://doi.org/10.52462/jlls.219
International Journal of Intercultural Relations, 35(1), 124–134.
26

Computers and Education Artificial Intelligence

Uploaded by

Copyright:

Available Formats

Computers and Education Artificial Intelligence

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Computers and Education Artificial Intelligence

Uploaded by

Copyright:

Available Formats

Computers and Education: Artificial Intelligence 4 (2023) 100134

Contents lists available at ScienceDirect