Implementing Artificial Intelligence for Communicative Arabic Language Learning in Indonesian Pesantren: A Theoretical Review

Abstract

This theoretical review investigates the implementation of artificial intelligence (AI) in enhancing communicative Arabic language learning for non-native speakers in Indonesia’s pesantren-based K-12 education. Grounded in the Communicative Language Teaching (CLT) framework, the study synthesizes literature from 2018–2025, focusing on AI tools such as natural language processing (NLP), intelligent tutoring systems (ITS), and conversational agents. The aim is to explore how these technologies align with CLT principles, particularly interaction, feedback, and learner autonomy, while considering the socio-cultural context of Islamic education in Indonesia. The study utilized an integrative theoretical review methodology, identifying and analyzing peer-reviewed publications from databases including ScienceDirect, Scopus, Web of Science, and Google Scholar. The findings reveal that AI-driven applications can provide adaptive, personalized language learning environments that enhance speaking, listening, and writing skills while reducing learner anxiety and increasing motivation. Specifically, AI tools offer real-time feedback, simulate interactive communication, and allow learners to engage with content in meaningful ways aligned with their cultural and religious context. The review concludes that AI holds substantial promise in transforming Arabic language instruction within pesantren settings, provided implementation is sensitive to cultural and infrastructural conditions. The paper contributes theoretically by linking AI functionalities with CLT pedagogical goals and practically by offering guidelines for educators and policymakers. It recommends pilot programs, teacher training, and context-aware content customization to ensure effective integration. The study highlights the need for further empirical research to assess long-term outcomes and optimize AI-assisted Arabic instruction for non-native speakers in Islamic educational settings.

Keywords: Artificial Intelligence; Communicative Language Teaching; Arabic as a Foreign Language; Pesantren Education; Intelligent Tutoring Systems; Natural Language Processing

Introduction

Arabic language proficiency is a core component of the curriculum in many Indonesian Islamic boarding schools (pesantren), valued both for religious literacy and for connecting learners to the wider Muslim world[1]. In the pesantren tradition, Arabic is often taught as a foreign language to non-native speakers (typically Indonesian youth) to enable them to read classical texts and participate in religious discourse. However, the 21st-century educational landscape demands that Arabic instruction evolve beyond rote learning of grammar and translation towards developing actual communication skills. Traditional pedagogy in many contexts, including some pesantren, has relied heavily on teacher-centered methods, memorization of vocabulary and scripture, and grammar-translation techniques. Research in developing educational systems has shown that such rote memorization and grammar-translation methods fail to foster the communicative competence needed for real-world language use[2]. When students are unable to use the language to speak, debate, create, or solve problems, their motivation for learning can plummet[3]. This issue is evident in Arabic language classes where students may excel at parsing texts or reciting rules, yet struggle to carry out a basic conversation in Arabic.

Communicative Language Teaching (CLT) offers a theoretical framework to address this gap. CLT emphasizes interaction, meaningful communication, and using the target language in authentic contexts as the primary goals of instruction. Rather than learning language as an abstract system, students learn by using the language to negotiate meaning, similar to how first languages are acquired. In the CLT approach, activities such as role-plays, discussions, and problem-solving tasks are prioritized over drills and rote translation. The aim is to develop learners’ communicative competence – the ability not only to apply grammatical rules but to know when and how to use language appropriately in various situations. CLT has gained global acceptance as an effective approach for modern language education, and is increasingly advocated for teaching Arabic as a second/foreign language as well[4]. In Indonesia, applying CLT in Arabic instruction aligns with national curriculum shifts towards student-centered and competency-based learning, though implementation in traditional pesantren settings can be challenging due to entrenched methods and limited resources.

At the same time, the rise of artificial intelligence in education offers new opportunities to reinforce communicative language learning. AI technologies – particularly those drawing on natural language processing (NLP) and machine learning – have advanced significantly in recent years and are being applied to language education in various forms. Intelligent Tutoring Systems (ITS) and AI-driven language learning applications can simulate aspects of one-on-one tutoring by providing immediate feedback, adaptive practice, and rich interactive experiences for learners. For example, AI-powered conversational agents (chatbots) can engage students in dialogue practice in the target language, offering a safe, judgment-free environment for learners to practice speaking without the anxiety of peer judgment[5]. Such chatbots are equipped with NLP capabilities to understand learner inputs (even if imperfect) and respond meaningfully, thereby enabling authentic communicative exchanges on a range of topics. Initial studies have suggested that these AI conversational partners can increase students’ willingness to communicate and reduce speaking anxiety in foreign language learning[6,7]. Similarly, intelligent language tutoring systems can tailor grammar and vocabulary exercises to a student’s proficiency level, track their errors, and provide targeted remediation, thus personalizing the learning experience in ways a single teacher with a large class might struggle to do.

Given these developments, it is pertinent to investigate how AI can be leveraged to enhance Arabic language learning in pesantren-based education, with a focus on improving communicative skills for non-native speakers. The pesantren context is unique: it blends traditional religious education with elements of modern curricula, and often operates with limited technological infrastructure. Many pesantren have historically been slow to adopt educational technology for classroom instruction – technology use might be confined to administrative tasks or basic computer classes, rather than language pedagogy[8]. However, there is growing recognition that modernizing Arabic teaching through communicative approaches and technology is crucial for preparing students for a digital, interconnected world[9][10]. Recent work in Aceh, Indonesia, for instance, proposed an integrative model that rejuvenates traditional Arabic instruction by incorporating CLT methods and context-aware digital tools, resulting in enhanced linguistic skills and learner motivation[9]. This reflects a broader call in the field for “melding tradition with innovation” so that Islamic education can flourish academically and spiritually in the contemporary era[11].

In light of the above, this study aims to analyze the implementation of AI in Arabic language learning for Indonesian K-12 students in pesantren, particularly to support communicative competence development. The guiding research questions include: (1) What types of AI-driven tools and applications have been used to facilitate foreign language learning (with an emphasis on Arabic and similar less-commonly-taught languages)? (2) How do these AI applications align with or support the principles of communicative language teaching, especially in improving speaking and interactive skills? (3) What are the theoretical and practical implications of integrating AI into Arabic language pedagogy in pesantren, and what challenges might need to be addressed in this context? By addressing these questions, the paper seeks to provide a comprehensive overview of current knowledge and offer a conceptual framework for educators interested in deploying AI to enhance Arabic learning. Ultimately, this work strives to contribute both to theory – by situating AI usage within CLT – and to practice – by offering evidence-based recommendations for implementation in Indonesian Islamic educational settings.

Methodology

Research Design

This study follows a theoretical review approach, aiming to analyze, synthesize, and integrate conceptual and empirical findings on the use of Artificial Intelligence (AI) in Arabic language learning through the lens of Communicative Language Teaching (CLT). A theoretical review differs from systematic or scoping reviews by emphasizing the construction of new theoretical frameworks and interpretations from existing literature, rather than aggregating data for empirical generalization.

The research process involved identifying scholarly works that contribute to theoretical insights into AI applications in language education, particularly in Arabic for non-native speakers, and their pedagogical implications for K-12 Islamic boarding schools (pesantren) in Indonesia. Sources were drawn from peer-reviewed journals, conference proceedings, and academic books available in databases such as ScienceDirect, Scopus, Web of Science, and Google Scholar between 2018 and 2025.

The selection process focused on literature that:

• Examined AI technologies such as NLP, intelligent tutoring systems, and chatbots in language learning contexts;
• Discussed communicative competence and the principles of CLT;
• Provided conceptual frameworks or pedagogical models applicable to Arabic or similar morphologically rich languages;
• Considered socio-cultural and institutional contexts relevant to pesantren or similar educational systems.

The collected works were evaluated qualitatively through conceptual mapping and thematic synthesis, identifying recurring constructs, assumptions, and models. Particular attention was paid to how AI tools align with CLT’s focus on meaningful interaction, feedback, and learner autonomy. Through iterative analysis, the review constructed a synthesized argument and proposed a conceptual model linking AI capabilities with communicative language pedagogy in pesantren settings.

No new empirical data were collected. Instead, this review draws theoretical generalizations and practical implications from existing knowledge to advance scholarly discussion and inform future empirical studies.

Data Sources and Collection

We systematically searched for relevant literature published primarily between 2018 and 2025 to capture the latest developments in AI-enhanced language learning. Key databases and indexing services were utilized, including ScienceDirect, Scopus, Google Scholar, and Web of Science (WoS). Search terms covered multiple facets of the topic, such as “artificial intelligence AND language learning”, “AI AND second language acquisition”, “intelligent tutoring system AND Arabic”, “chatbot AND language education”, “communicative language teaching AND technology”, and “Arabic language learning AND Indonesia”. We focused on peer-reviewed journal articles, conference proceedings, and reputable reports. Given the interdisciplinary nature of the subject, relevant studies were drawn from fields of applied linguistics (e.g. computer-assisted language learning), educational technology, and computer science (particularly NLP). To ensure high quality and relevance, preference was given to sources indexed in Scopus or Web of Science, as well as those in established language learning journals and AI-in-education conferences.

The inclusion criteria for literature were: (a) studies or reviews that specifically examine the use of AI or AI-driven tools in language learning or teaching; (b) among these, priority to works focusing on communicative skills (speaking, conversation, interactive writing) or overall language proficiency outcomes; (c) evidence from K-12 education settings or young learners when available (since pesantren generally cater to school-age students, roughly equivalent to secondary education); and (d) publications that discuss Arabic language learning or contexts similar to it (e.g. other less commonly taught languages, or settings in the Muslim world) to draw parallels. Additionally, theoretical papers on CLT and technology integration were included to frame the pedagogical perspective. Some Indonesian national or regional studies were also reviewed to capture context-specific insights, such as technology adoption in pesantren and Arabic teaching methodologies in Indonesia, even if these sources were not indexed internationally.

Analysis Procedure

We conducted a thematic analysis of the collected literature. Each source was reviewed for key findings related to AI applications (types of tools, features), reported impacts on learning (skill improvements, motivational or affective outcomes), and any noted challenges or recommendations. These points were coded into themes such as “AI for speaking practice”, “intelligent tutoring/adaptive learning”, “learner engagement and motivation”, “communicative competence outcomes”, and “implementation challenges”. We then examined how these themes intersect with Communicative Language Teaching (CLT) principles. For instance, evidence of AI enabling interactive dialogue was mapped to CLT’s emphasis on meaningful communication; instances of personalized feedback were related to the concept of scaffolding within a learner’s zone of proximal development (a concept from socio-constructivist theory complementary to CLT).

To integrate the pesantren case context, we gathered descriptive information on the current state of Arabic education in pesantren from literature and official reports. This included typical teaching methods, class sizes, teacher competencies, technological infrastructure, and cultural factors. We then used an analytic framework to consider how the general findings on AI and language learning might translate into this setting. This involved asking questions during analysis, such as: Would this AI tool address a specific need or challenge in pesantren Arabic classes?; What modifications or supports would be needed for it to work in a pesantren environment?; Does implementation align with the values and goals of pesantren (for example, respecting Islamic educational values while introducing modern technology)? The discussion section of this paper synthesizes these analytical considerations.

No human subjects were directly involved in this research; it is a literature-based study. Therefore, issues of ethics approval or data privacy were not applicable beyond the ethical use of sources. However, in interpreting results, we remained mindful of potential biases in literature (such as an over-representation of studies on English teaching, or positive publication bias in reporting AI benefits) and we note these limitations where relevant. By combining evidence from prior studies with context-specific reasoning, the methodology aims to build a well-founded argument for the potential of AI in communicative Arabic language teaching, as well as an honest appraisal of what gaps remain.

Results

AI Applications in Language Learning: NLP and Intelligent Tutoring Systems

The review identified several key categories of AI-driven applications being used in language education. Foremost among these are intelligent tutoring systems (ITS) for language learning and conversational agents (chatbots) that leverage natural language processing. Intelligent tutoring systems are computer programs designed to provide individualized instruction and feedback, mimicking a human tutor’s guidance. In the language learning domain, ITS often incorporate expert knowledge of language (e.g. grammar rules, vocabulary) and student modeling to adapt to a learner’s progress. A notable example is Arabic Grammar Tutor (AG_Tutor), an intelligent tutoring system developed to teach Arabic grammar to elementary students[12]. AG_Tutor includes modules for delivering tutorial content, selecting appropriate exercises, and analyzing student responses with an expert module[12]. The system was built around the fourth-grade Arabic grammar curriculum in Egypt and employed NLP techniques (such as morphological analysis of Arabic words) to evaluate answers and provide feedback. This illustrates how AI can handle the rich morphology of Arabic, a language that poses significant challenges to NLP due to its complex word structures and ambiguity[13]. By integrating an Arabic morphological analyzer and grammar rules into the ITS, the tutor can automatically detect specific errors (for instance, incorrect verb conjugation or case endings) and give tailored hints or explanations. Early testing of such systems showed that they can effectively simulate a one-on-one tutoring experience, with the potential to improve learners’ mastery of difficult grammar concepts by allowing practice at one’s own pace in a low-pressure environment.

Another class of AI applications in language learning is conversational AI agents or chatbots. These are programs that can engage in written or spoken dialogue with learners in the target language. Modern language-learning chatbots use advanced NLP (often powered by machine learning or even large language models) to understand learner inputs and generate contextually appropriate responses. Importantly, they are typically designed to handle unstructured input – meaning a student can say or type something in the target language, and the AI will attempt to interpret the intent and respond meaningfully, rather than following a strict script. This opens up possibilities for open-ended communicative practice, closely aligning with CLT’s focus on authentic communication. Several studies have explored chatbots as conversation partners for language learners. For example, Yang et al. (2022) implemented an AI chatbot as a virtual English conversation partner in EFL classes and found it improved students’ willingness to communicate while reducing their speaking anxiety[6]. The chatbot provided a human-like interlocutor that was available anytime for practice, and students reported feeling more comfortable making mistakes or trying new expressions with the AI than they might be with a human teacher or peers (since the AI is seen as non-judgmental). Similarly, Hwang et al. (2024) and others have reported that AI-driven conversational practice can enhance learners’ fluency and confidence by offering plentiful interactive speaking opportunities in a safe environment[14]. These chatbots can be configured for Arabic as well – indeed, emerging commercial platforms (e.g. LanguaTalk, TalkPal, ArabicTutorAI) specifically offer AI chatbots for Arabic conversation practice, indicating the technology is becoming language-agnostic and can support less commonly taught languages like Arabic[15][16].

Closely related to chatbots are AI-powered intelligent language tutoring systems with dialogue. Some advanced systems integrate tutoring strategies with conversational ability – for instance, an AI might guide a learner through a role-play scenario in Arabic, asking questions and providing prompts, then giving feedback on the learner’s responses. This merges the idea of an ITS (providing feedback and hints) with interactive conversation. A specialized example is LANA, an Arabic conversational ITS developed to help children (including those with special needs) practice social dialogues; it uses multiple AI agents to manage conversation flow and feedback, demonstrating the feasibility of AI-mediated Arabic speaking practice even for young learners with varying needs[17]. Although LANA was tailored for a specific context (children with Autism Spectrum Disorder), its development showcases that Arabic conversational tutoring by AI is technically achievable and can be designed to adapt to learner responses in real time.

In addition to full-fledged tutors and chatbots, AI is also embedded in more narrow tools that assist language learning. Speech recognition technology powered by AI is increasingly used for pronunciation training and assessment. For instance, AI-based speech recognition can evaluate a student’s spoken Arabic, detect pronunciation errors or mispronounced phonemes, and provide instant feedback or corrective modeling. This is especially relevant for Arabic, which contains several sounds that are unfamiliar to Indonesian learners (such as certain emphatic consonants or guttural sounds). By using speech recognition, an application can pinpoint which syllables the learner is struggling with and perhaps highlight the difference between the learner’s pronunciation and a native-like pronunciation. Research on English learners has shown that AI-powered speech recognition tools can significantly improve pronunciation accuracy and speaking intelligibility[18]. In one study, integrating an intelligent voice assistant in speaking practice led to notable gains; for example, Arabic native speakers learning English experienced a 34.2% reduction in pronunciation errors after practicing with the AI assistant[19]. Although that particular study was on Arabic speakers learning English, the underlying principle – that targeted feedback via AI can enhance pronunciation – can be applied conversely to Indonesian speakers learning Arabic. The ability of AI to give immediate, private feedback on speech could help pesantren students practice difficult Arabic sounds more frequently without the fear of embarrassment.

AI-driven writing assistants form another category. These include automated essay scoring systems and grammar/spell checkers that are far more sophisticated than earlier generations of software. Using NLP, such tools can not only flag grammatical errors but also make stylistic suggestions, check for appropriate use of expressions, and even assess semantic coherence to some degree. While much of the development in this area has been for English, there are tools emerging for Arabic writing as well (thanks to progress in Arabic NLP). For example, an AI system can analyze an Arabic essay and highlight subject-verb agreement errors or improper usage of vocabulary, providing explanations in Arabic or the learner’s first language. This functions like a virtual writing tutor. By receiving detailed feedback on their writing, students can learn from mistakes and improve their ability to express ideas in written Arabic. Additionally, machine translation (MT), an AI-powered tool, has been co-opted as a learning aid: students can use MT (like Google Translate or specialized Arabic-English translators) to draft ideas and then refine the output, or to check their own writing. Studies on machine-translation-assisted language learning suggest that, when used judiciously, MT can help learners notice gaps in their knowledge and expand vocabulary, though it should be combined with guidance to avoid overreliance[20]. It’s worth noting that MT for Arabic has improved over time but still can produce errors due to Arabic’s linguistic complexity; this can itself become a learning point if students critically evaluate the MT output.

In summary, AI applications in language learning range from tutoring systems that adapt to individual learners, to conversational AI partners for dialogue practice, to NLP-based tools for specific skills like pronunciation and writing. All these tools share an underlying strength: they utilize AI’s capacity for automation, adaptivity, and analysis of natural language to provide learning experiences that would be difficult to scale with human teachers alone. (Table 1 – hypothetical – is omitted here, but it would summarize these AI tools and their functions in the context of language education.) The next sections discuss the observed impacts of these AI tools on learning outcomes, particularly focusing on communicative competence and affective factors, as reported in recent studies.

Impact of AI Integration on Language Learning Outcomes

A dominant theme in the literature is that integrating AI-driven tools into language instruction tends to have positive effects on student learning outcomes across various language skills. Notably, speaking and listening skills – which are central to communicative competence – see significant improvements when learners engage with AI-enabled practice. A meta-analysis by Qiao and Zhao (2023) investigated AI-based instruction for Chinese EFL students and found significant gains in multiple aspects of speaking performance[21]. In their study, students who used an AI platform that provided interactive speaking practice and personalized feedback demonstrated higher post-test scores in fluency, vocabulary range, and accuracy of speech compared to a control group[21]. These findings align with other studies in the field. For instance, an experiment with university EFL learners in Indonesia showed that after six weeks of practice with an AI speaking application, students not only improved their objective speaking test scores but also reported greater self-confidence in speaking; importantly, there was a statistically significant reduction in their speaking anxiety levels as measured by questionnaires[22][23]. The reduction of foreign language speaking anxiety (FLSA) is a crucial outcome, because anxiety can inhibit learners from participating in communicative activities. AI tools create a private practice space where mistakes carry no social consequence, thereby lowering the affective filter and encouraging more frequent practice – a phenomenon consistently observed in studies where AI integration led to decreased anxiety and increased willingness to communicate[6][7].

Beyond speaking, AI has also been found to aid other language competencies. Vocabulary acquisition, for example, can be enhanced through AI-driven personalized review systems. Adaptive flashcard apps or word-learning games can employ machine learning to identify which words a student struggles with and schedule them for review (spaced repetition) at optimal intervals. A recent systematic review highlighted that AI-driven systems contribute to better vocabulary retention compared to traditional methods[24]. The use of adaptive algorithms ensures that learners get more practice on words they frequently forget, and less on words they already know, thereby improving long-term retention. Additionally, some AI applications incorporate semantic analysis to teach vocabulary in context – for instance, a reading tutor that can explain new words in a text or a chatbot that dynamically simplifies its vocabulary if the learner is not understanding, then gradually re-introduces more difficult words as the learner’s proficiency improves[5][14].

Listening comprehension is another skill that can benefit from AI, although it is less directly targeted by interactive AI (since most chatbots focus on reading/writing or spoken conversation where listening is inherent). AI can aid listening practice by generating endless variations of spoken dialogues or questions (text-to-speech systems can voice the chatbot’s responses in Arabic, providing exposure to spoken language). Moreover, AI systems can analyze a learner’s responses to listening tasks – for example, if a learner answers a listening comprehension question incorrectly, an intelligent system might pinpoint whether the mistake was due to a specific linguistic feature (like misunderstanding a negation or a time reference) and then provide focused listening practice on that feature. However, robust evidence on AI’s impact solely on listening skill is still emerging, often tied with speaking improvements since the two are practiced together in conversation.

Reading and writing skills also show positive outcomes with AI support. In writing, students using AI writing assistants have shown improvements in grammatical accuracy and text organization over the course of a semester, as the tool flags recurring errors and offers suggestions. One study using an AI-enhanced digital storytelling activity (students created video dramas with AI support for scripting and language correction) found that participants improved both their writing and speaking abilities by the end of the project[25]. The AI’s guidance in drafting scripts and practicing pronunciation for the video drama provided a form of integrated skills training that reflected real-world communicative use of language. In reading, AI can tailor text difficulty to a learner’s level – for example, an AI reading platform might simplify passages or provide instant dictionary help. There is evidence that such AI-enhanced reading tools can increase reading comprehension and even track biometric feedback (like gaze or pause durations) to adjust the reading experience[26]. (Though specific references on Arabic reading are scant, it is reasonable to extrapolate from general L2 reading research.)

Importantly, apart from linguistic skills, affective and cognitive factors are positively influenced by AI integration. Several studies note heightened learner motivation and engagement when AI tools are part of the learning process[24]. The interactive and game-like nature of many AI applications (such as earning points in a tutoring system or the novelty of chatting with a “robot”) can sustain interest. Moreover, the personalized nature of AI feedback helps learners set and achieve personal goals, contributing to a sense of autonomy and competence which are known motivators in self-determined learning. AI tutors can also foster better self-regulated learning behaviors – for example, an AI platform might prompt learners to practice regularly, track their own progress in dashboards, and identify areas for improvement, thereby encouraging metacognitive planning and reflection. In Qiao & Zhao’s (2023) study, the group using AI showed greater self-regulation in speaking practice compared to the traditional group, as the immediate feedback loop and progress indicators encouraged students to take charge of their learning[21]. Such findings are echoed by others, suggesting AI’s capacity to not only teach content but also to coach learners in effective learning strategies.

To summarize the impact: AI integration in language learning has demonstrated benefits such as improved speaking fluency and accuracy, increased vocabulary retention, better writing quality, reduced anxiety, and higher motivation[27][24]. These outcomes have been observed in various contexts (mostly with English as the target language, but increasingly in other languages as well). While research specifically on Arabic language learners is still limited compared to English or other major languages, the general trends are likely applicable, given the human learning processes are similar and the AI tools can be language-agnostic if properly trained or configured for Arabic. Next, we delve deeper into how AI tools can specifically support the development of communicative competence – the primary goal of CLT – with a focus on the oral communication needs of Arabic learners in pesantren.

AI for Communicative Competence in Arabic: Speaking Practice and Interaction

One of the most salient contributions of AI to language pedagogy, particularly under a communicative paradigm, is the expansion of opportunities for meaningful practice in speaking and interacting. In a traditional classroom, especially in many Asian contexts, students often get limited chances to speak the target language due to large class sizes or teacher-centered activities. This is an area where AI can significantly augment learning: an AI conversational partner can engage each learner in dialogue simultaneously, something a single teacher cannot do with dozens of students at once. For Arabic language learners in pesantren, practicing speaking has unique challenges. Many pesantren place heavy emphasis on reading and translating classical Arabic texts (for religious studies), and less on everyday conversational Arabic. Students might learn formal grammar and a literary vocabulary, but not how to ask for directions, describe their daily routine, or discuss a contemporary topic in Arabic. This is a gap where communicative practice is needed. AI chatbots or voice-based assistants designed for Arabic can simulate conversations on practical topics – for example, a chatbot could role-play a scenario like shopping in a market in an Arabic-speaking country, or having a casual chat about hobbies. By doing so, it exposes students to the kind of interactive use of Arabic that mirrors real-life communication, thus operationalizing the principles of CLT.

In simpler terms, the chatbot can adapt its level of language to be just above the learner’s current level, ensuring that the interaction is challenging yet comprehensible – a key condition for language development under both CLT and broader interactionist theories. For instance, if a learner is struggling, the AI can switch to simpler vocabulary or shorter sentences; if the learner is excelling, the AI can introduce new colloquial phrases or ask open-ended questions to push the learner to expand their output[14]. This dynamic adjustment is powered by NLP algorithms that analyze the learner’s input in real time. The net effect is a form of personalized conversational scaffolding: the AI helps learners produce language they might not manage alone by providing timely prompts, rephrasing questions, or giving feedback – much like a skilled teacher would do in a one-on-one conversational tutoring session[30][31].

The ability of AI to give real-time feedback in speaking practice is especially valuable. In human conversation, correcting every mistake is neither feasible nor desirable (it would interrupt the flow), and teachers usually cannot correct all errors during a communicative activity. An AI, however, can be designed to unobtrusively note errors and address them either immediately or later in the dialogue. For example, if a student consistently uses the wrong verb form, the chatbot might respond with a sentence that recasts the student’s error in corrected form, thereby modeling the right usage without halting the conversation. Some systems also have a “feedback mode” where after a conversation, the learner can ask the AI, “How did I do?” and receive a summary of mistakes or suggestions. Studies have noted that such immediate and detailed feedback helps learners become aware of their gaps and correct them, reinforcing learning much faster than waiting for a teacher to mark errors in a test or homework[32][21]. In essence, AI provides an on-demand tutor for communicative practice, available whenever the learner has time to converse. This is highly relevant for pesantren students, who often have rigorous schedules; an AI that can chat during evening self-study periods or after classes could give additional speaking practice beyond the classroom curriculum.

Communicative competence is not only about linguistic skills but also about confidence and willingness to communicate (WTC) in the target language. It has been observed that repeated successful interactions with AI can boost learners’ confidence. One empirical example: after practicing with an AI chatbot, EFL students in one study showed improved WTC and a positive shift in attitudes, feeling less afraid to speak English in front of others[6]. Although that study was in the context of English, the psychological effect would likely be similar for Arabic. In pesantren, where the classroom culture might be quite formal and respectful of teacher authority (leading students to be shy in speaking out loud), the AI offers a low-stakes venue for voice. Over time, as students get used to formulating thoughts in Arabic with the AI, they can transfer that confidence to real conversations with teachers, peers, or native speakers. Indeed, some pesantren have “Arabic days” or language immersion policies – AI tools could reinforce these by ensuring that even outside those designated times, students have the means to practice communicatively.

It is also crucial to mention that AI can expose students to diverse communicative scenarios and dialects. Modern standard Arabic (Fusha) is taught in schools, but Arabic as used in real communication often involves dialects or at least a less formal register. AI systems could, for instance, simulate a speaker from Egypt vs. one from the Gulf, helping students attune to variations in pronunciation or common colloquialisms (assuming the system is trained on those dialectal corpora). This breadth of exposure is something even a very skilled teacher cannot easily provide, but an AI could incorporate multiple language models or personas. Having said that, designing AI to handle dialectal Arabic is complex and still developing, but it represents a future direction that could greatly enhance communicative competence training for learners aiming to use Arabic in international contexts.

In summary, AI tools — particularly conversational agents — richly support the communicative aims of language teaching by providing interactive practice, adaptive scaffolding, immediate feedback, and confidence-building opportunities. These align closely with CLT’s core tenets: learners are actively using language to negotiate meaning, the focus is on conveying and understanding messages (even as form is corrected in the background), and the context of language use is varied and meaningful. The next section will consider how these general benefits and tools translate into the specific environment of Indonesian pesantren, including potential challenges to implementation.

Contextualizing to Pesantren Education: Opportunities and Challenges

Integrating AI-driven communicative language learning in pesantren-based education in Indonesia presents both exciting opportunities and particular challenges. Pesantren, being Islamic boarding schools, have a distinct educational climate. They combine religious instruction with general education, and Arabic typically holds a dual role: it is a subject of study (as a language) and a tool for religious learning (many Islamic texts are in Arabic). There is often a deep respect for traditional methods – for instance, time-honored approaches like sorogan (one-on-one reading with a teacher) or bandongan (teacher reading texts aloud while students listen and annotate) are still practiced for religious studies. Introducing AI into this mix must therefore be done in a way that complements and enhances the educational values, rather than appearing to sideline them.

Opportunities: The findings from this review suggest several ways AI could be leveraged in pesantren to improve Arabic communicative competence:

• Supplementing Classroom Instruction: AI tutoring systems or chatbots can act as supplementary tools for students to practice Arabic outside of class hours. For example, after a lesson on daily conversation, students could be assigned to practice with an Arabic chatbot in the computer lab or on their mobile devices (if allowed) to reinforce what they learned. This extends learning beyond the limited classroom speaking practice. Given that pesantren students live on campus, structured time could be allotted in the evening for such AI-mediated practice, essentially adding a “virtual language lab” component to their routine.
• Differentiated Learning: In a typical pesantren Arabic class, students’ proficiency may vary widely – some might come from backgrounds with more exposure to Arabic (e.g., from modern Islamic schools) while others are beginners. AI systems shine in offering differentiated instruction; an intelligent tutor could adapt tasks to each student’s level. One student might get more basic vocabulary exercises while another gets complex sentence construction tasks, all under the same system but personalized. This addresses the heterogeneity of learner levels which teachers often struggle with in large classes.
• Task-Based Communicative Projects: AI can enable innovative project-based learning in Arabic. For instance, students could work in groups to create a short Arabic dialogue video, using AI for help: an AI assistant might help them check their script or even provide an initial script that students then modify. Another idea is using AI to connect pesantren students with Arabic speakers globally via mediated chat – some platforms use AI to facilitate tandem learning, where half the time is spent in one language and half in the other, with AI ensuring fairness and providing translations when needed. While these are beyond simple chatbot use, they illustrate how AI can facilitate authentic communicative tasks even in a relatively secluded school environment.
• Pronunciation and Recitation: Given the importance of recitation in pesantren (for Qur’anic Arabic especially), AI pronunciation tools could be very valuable. They could be used not only to correct pronunciation in spoken Arabic communication but also to help with tajwīd (rules of Qur’anic recitation) by detecting errors in articulation points of letters. This dual utility (for secular language learning and for religious recitation practice) might make stakeholders more receptive to adopting the technology, as it directly supports core religious activities.

Challenges: Despite the potential, several challenges must be considered:

• Infrastructure Constraints: Many pesantren, especially in rural areas, may have limited internet access, outdated computer hardware, or not enough devices for students. AI tools, especially those requiring online processing (like cloud-based NLP chatbots), need reliable internet and devices. Addressing this requires investment in ICT infrastructure at pesantren – from providing broadband connections to setting up computer labs or allowing controlled use of smartphones/tablets. As noted in recent policy recommendations, infrastructure development and access democratization are crucial for technology-enhanced learning in pesantren[33].
• Teacher Training and Acceptance: Teachers in pesantren might not be familiar with AI tools and could feel apprehensive about using them. There might be a perception that technology could replace the teacher’s role or that it does not align with the traditional teaching of sacred subjects. It is vital to design teacher professional development programs focusing on how AI tools can be used to augment teaching, not replace it[33]. Workshops could train Arabic teachers to use specific applications, interpret AI-generated feedback, and integrate AI activities into lesson plans in a pedagogically sound way. Having teachers on board is key – if they see AI as a helpful assistant (for example, helping weaker students catch up or handling routine practice so the teacher can focus on higher-order teaching), they are more likely to embrace it.
• Content Alignment and Cultural Relevance: AI systems need to be tuned to the curriculum and cultural context of the learners. For Arabic in pesantren, this might mean ensuring the AI’s vocabulary and topics include Islamic and Indonesian cultural contexts. For example, dialogues about daily life should be relatable to Indonesian students (talking about local foods or customs, perhaps), and any sensitive content must be controlled (a pesantren would avoid material that is not in line with Islamic values). Developing or selecting AI tools that allow content customization is therefore important. If a generic chatbot is used, it should be reviewed for appropriateness. There is ongoing research on localizing AI content to various cultures; in this case, collaboration with Islamic education experts could guide the content of AI-driven tasks to ensure they are respectful and relevant.
• Arabic NLP Limitations: While AI for language learning has progressed, Arabic NLP still lags behind English in some respects. Arabic’s complexity means that off-the-shelf language models might make errors in understanding or generating Arabic, especially for learners’ input which can be ungrammatical. This could lead to misunderstandings or the AI providing incorrect feedback. Thus, technical refinement is needed – either through improving Arabic NLP engines or constraining the interaction to what the AI can handle reliably. Ongoing advancements (e.g., large language models trained on Arabic data, improved speech recognition for Arabic with diverse accents) are mitigating this, but it remains a consideration. Collaborative efforts, possibly involving Indonesian universities or tech companies, could work on tailored AI solutions for Indonesian-accented Arabic or for the specific syllabus used in schools.
• Policy and Ethical Considerations: On a policy level, the Ministry of Religious Affairs (which often oversees pesantren education) would need to endorse and perhaps regulate the use of AI in classrooms. Issues of data privacy (especially if students’ voices or performance data are recorded by AI systems), screen time concerns, and equity (ensuring all students benefit, not only those in better-funded pesantren) should be addressed. Ethically, transparency about AI’s capabilities and limits should be communicated to both teachers and students to avoid over-reliance or misconceptions (e.g., students should understand that AI might occasionally give wrong advice and that teachers remain the ultimate authority on language matters).

Despite these challenges, the trajectory for integrating AI in pesantren is promising. The modernization of Arabic teaching is increasingly seen as necessary for students to become competent global communicators without losing the essence of their religious education[34]. By carefully aligning AI tools with pedagogical goals and cultural values, pesantren can innovate while preserving their identity – viewing technology as a means of cultural stewardship rather than a threat to tradition[34]. The following Discussion section will further interpret these findings and propose strategies to effectively harness AI for communicative Arabic language learning in this unique context.

Discussion

The convergence of artificial intelligence and communicative language teaching (CLT) in the context of Arabic education for non-native speakers presents a multifaceted scenario. In this section, we interpret the results through the dual lenses of pedagogy and technology, highlighting theoretical insights and practical recommendations. We also examine how the findings contribute to the broader discourse on language education and what they imply for stakeholders in the Indonesian pesantren system.

Theoretical Implications: AI as a Facilitator of Communicative Language Teaching

From a theoretical standpoint, the integration of AI tools in language learning reinforces and extends the principles of communicative language teaching. CLT posits that interaction in the target language, particularly in meaningful contexts, is essential for developing communicative competence. However, ensuring ample interaction has always been a pedagogical challenge, especially in foreign language contexts where exposure to native speakers or real-life situations is limited. AI addresses this gap by providing an interactive medium that can supplement human interaction. In essence, AI becomes a pedagogical agent that embodies CLT principles: it encourages learners to use language for genuine communication (e.g., discussing everyday topics with a chatbot), it often centers on conveying meaning rather than explicit grammar instruction (errors are corrected in passing rather than becoming the focus, mirroring how a communicative class handles form-focus), and it can simulate the unpredictability and creativity of real communication better than scripted classroom drills.

One theoretical insight is viewing AI-driven tools as extensions of Vygotsky’s concept of the More Knowledgeable Other (MKO) in socio-constructivist theory[28][29]. In traditional CLT classrooms, the MKO could be the teacher or a more proficient peer who scaffolds the learner’s attempts at using language. AI chatbots, as the results show, can act as virtual MKOs – guiding the learner within their Zone of Proximal Development by adjusting the difficulty of language and providing prompts[30]. This suggests a theoretical model where human-AI interaction is a valid form of social interaction for language learning, a notion that might have seemed far-fetched decades ago when CLT was conceived in entirely human terms. The positive outcomes (improved fluency, willingness to communicate, etc.) support the idea that interaction does not strictly require another human, as long as the essential elements of negotiation of meaning and feedback are present. This prompts a re-examination of interactionist theories in second language acquisition (SLA): AI interlocutors can provide confirmation checks, clarification requests, and other interactional moves that drive language development, which aligns with Long’s Interaction Hypothesis and related theories – albeit with AI in lieu of a human conversational partner.

The use of intelligent tutoring systems also intersects with cognitive theories of SLA, such as the Noticing Hypothesis and Skill Acquisition Theory. An ITS that gives detailed feedback helps learners notice the gap between their production and the target norm (Schmidt’s Noticing Hypothesis), as it explicitly points out errors or provides the correct forms. Over time, through repeated practice with adaptive difficulty (e.g., gradually increasing complexity of exercises as competence grows), learners proceduralize their knowledge – a process described by Skill Acquisition Theory. Traditionally, balancing the focus on form and meaning is a delicate matter for CLT practitioners (too much focus on form can undermine communicative practice, but too little can fossilize errors). AI offers a solution by handling a lot of form-focused feedback in the background or in individualized sessions, freeing classroom time for communicative use. Theoretically, this combination could lead to a more robust language competence: communicative confidence coupled with accuracy, as suggested by the improvement in both fluency and accuracy in studies[27][21]. It supports the notion that accuracy and fluency need not be a zero-sum trade-off if properly scaffolded; AI can help maintain that balance by providing immediate form feedback without derailing communication practice.

For communicative language teaching theory specifically in the Arabic-as-foreign-language context, this review provides a conceptual advancement. Arabic language pedagogy has often leaned towards grammar-translation and audiolingual practices historically, especially in religious education settings. Embracing CLT with the help of AI might serve as a catalyst to shift these pedagogical paradigms. The success of AI tools in fostering real communication capabilities (as seen in other languages) gives theoretical and empirical weight to calls for communicative approaches in Arabic. It’s one thing to argue for CLT in Arabic on principle; it’s another to demonstrate that learners can acquire communicative skills effectively when given the right practice environment, which AI now helps provide. Thus, the theoretical contribution here is highlighting AI’s role as an enabler of CLT in environments where traditional methodologies are deeply ingrained – offering a pathway to blend modern language acquisition theory with long-established educational frameworks.

Practical Implications and Recommendations

Practically, the findings from this study suggest a roadmap for educators and policymakers to integrate AI in Arabic language programs, particularly in pesantren schools. Below, we outline key recommendations and considerations:

• Pilot AI-Assisted Language Programs: It is advisable to start with pilot projects in a few willing pesantren to test how AI tools work in practice. For example, a pesantren could implement an AI chatbot for Arabic conversation practice over a semester with one class, while a comparable class continues with traditional methods, to evaluate differences in outcomes. If pilots (possibly done in collaboration with university researchers) show positive results in terms of student engagement and proficiency gains, it would build the case for scaling up. Early successes can create buy-in from the broader community of educators and administrators.
• Infrastructure and Resources: Ensure the necessary infrastructure is in place. This may involve setting up a computer lab or a tablet lending program if personal devices are not common. Where internet connectivity is an issue, consider AI solutions that can run offline or on local networks. For instance, certain speech recognition and tutoring software can be installed locally. If budgeting is a concern, seeking partnerships with ed-tech companies or government grants focused on digital education can be beneficial. Public–private partnerships, as suggested by recent studies, can help democratize access and support innovation in educational technology[33]. For example, a tech company might provide an Arabic learning app free of charge to pilot schools to showcase its effectiveness.
• Teacher Training and Involvement: Teachers should be brought on board early. Training sessions not only on how to use the AI tools but also on how to integrate them pedagogically are crucial. For instance, teachers can learn how to review chatbot logs to identify common student mistakes and address them in class (thus blending AI-driven learning with teacher-led follow-up). Teachers can also be trained to interpret AI analytics dashboards (many platforms provide data on student performance) to personalize their in-class instruction. Emphasize in training that the teacher’s role evolves into a facilitator and mentor in an AI-enriched environment – guiding discussions about chatbot conversations, clarifying any incorrect feedback the AI might have given, and ensuring students reflect on their learning. This will help mitigate any fears that AI is replacing the teacher; instead, it is a teaching aid that can handle repetitive tasks, allowing the teacher to focus on higher-level mentoring.
• Curriculum Integration: AI tools should be integrated into the curriculum rather than seen as extracurricular. This means aligning AI-based activities with learning objectives. For example, if a unit’s goal is to practice daily conversation, the curriculum can specify that students will engage in a 30-minute chatbot conversation as one of the practice activities. If the goal is mastering a grammar point, an ITS exercise could be assigned. Integration also means assessment can include components that account for AI-based practice (perhaps via a portfolio where students submit AI interaction summaries or improved scores from an AI tutor over time). This makes AI a part of the learning process that counts, rather than a gimmick or optional tool.
• Content and Context Customization: Choose or customize AI content to fit the pesantren context. If using a commercial AI app, work with the provider to include context-appropriate content (for instance, dialogues set in an Islamic school, vocabulary related to daily activities in a pesantren, or topics that interest these students). Alternatively, use more open platforms (some AI chatbot frameworks are open-source) to develop your own conversational scripts and knowledge base focusing on the desired content. The communicative tasks given by the AI should resonate with students’ lives or aspirations – e.g., discussing plans for Eid holidays in Arabic, or talking about one’s family – which makes the practice meaningful and culturally relevant.
• Monitoring and Evaluation: Continuously monitor both the learning outcomes and the process. Collect feedback from students and teachers about their experiences with the AI tools. Are students finding it engaging or frustrating? Are teachers noticing improvement in class participation or language use? Quantitatively, compare test results or proficiency assessments before and after AI implementation, keeping in mind that standardized testing of speaking could be utilized (possibly even leveraging AI assessment). This data will be important to adjust the approach and also to convince stakeholders of the value added. It will also highlight any issues (e.g., if students are gaming the system, or if technical glitches are hindering use).
• Addressing Challenges Proactively: Anticipate and address the challenges noted. For infrastructure, have contingency plans (like offline backups for when internet fails). For teacher acceptance, perhaps involve religious authorities or respected figures to endorse the approach, framing it as enhancing the understanding of Arabic which in turn allows better understanding of religious texts. For example, an argument might be made that by learning to communicate in Arabic, students can more deeply internalize the language of the Qur’an in context, rather than just memorizing it – a perspective that can be appealing in Islamic education. On the technical side, maintain an updated system – if an AI tool is not performing well in understanding student input, gather those cases and work with developers to improve the system’s accuracy for those language patterns.

Overall, the practical steps revolve around strategic implementation, collaboration, and continuous adaptation. When executed thoughtfully, the introduction of AI in Arabic learning within pesantren could become a model for blending traditional and modern pedagogies. It can show that embracing technology does not equate to westernizing or secularizing the education, but rather can strengthen students’ linguistic capabilities to engage both with their heritage and with the modern world.

Limitations and Future Research

While this review makes a strong case for AI’s potential in Arabic language education, it is important to acknowledge its limitations. First, the majority of empirical studies available are on English or other more commonly taught languages; direct research on AI in Arabic learning (especially in K-12 or pesantren contexts) is still sparse. Therefore, some extrapolation was used in our analysis. Future empirical research should be conducted in actual pesantren classrooms to validate the effectiveness of specific AI interventions for Arabic – for example, a controlled study measuring speaking proficiency gains from chatbot practice in an Indonesian Arabic class would provide concrete evidence to support (or refine) our claims.

Secondly, the rapid pace of AI development means that tools available today might be outdated soon. We focused on current literature up to 2025; however, new large language models or AI platforms (potentially including widely available ones like ChatGPT and its successors) are continually improving in conversational ability. Future work could explore how the latest generative AI models, which are capable of very human-like dialogue, might further revolutionize language practice – and what new challenges (bias, over-reliance, cost) they bring.

Another area for future research is the long-term impact of AI integration. Most studies measure immediate gains in skills or short-term course outcomes. It would be valuable to track students over several years to see if early exposure to AI-enhanced learning leads to higher ultimate proficiency or sustained language use. For instance, do pesantren alumni who learned with AI exhibit better communicative Arabic skills in university or the workplace? Longitudinal studies or at least follow-up surveys could address this.

Finally, research should also delve into the affective and social dimensions more deeply: How do students perceive learning with AI? Does it increase their enjoyment (some evidence says yes, through foreign language enjoyment metrics[35])? Does it change the classroom social dynamics in any way? Understanding these human factors will guide how teachers orchestrate the blend of AI and traditional interaction, ensuring that technology remains a positive force in the learning ecosystem.

Conclusion

This study set out to examine the implementation of artificial intelligence for communicative Arabic language learning in Indonesia’s pesantren-based K-12 education, through a comprehensive review of literature and theoretical analysis. The findings affirm that AI technologies – from intelligent tutoring systems that adapt to individual learners, to conversational chatbots that provide interactive speaking practice – offer powerful tools to enhance language acquisition in alignment with communicative language teaching principles. AI-driven interventions have demonstrated the ability to improve key language skills (speaking fluency, accuracy, vocabulary, and writing competence) and beneficially influence affective factors (reducing speaking anxiety and boosting motivation and confidence) in second language learning[27][7]. These outcomes are highly pertinent to the Arabic learning context in pesantren, where developing real communicative ability has historically been a challenge under more traditional pedagogies.

By situating AI within the CLT framework, we highlighted that technology can facilitate authentic communication and provide the kind of individualized, meaningful practice that students need to become competent language users. The theoretical contribution of this work lies in bridging modern AI-enabled educational practices with established language teaching theory – demonstrating that concepts like scaffolding, interaction, and learner-centered instruction can be effectively operationalized through AI tools. In the case of Arabic, a language often taught for its liturgical importance, this fusion of AI and CLT could help transform learning from a passive reception of rules into an active skill-building process where students truly learn to use Arabic in context. Moreover, our focus on the pesantren context adds a culturally nuanced perspective, showing that innovation in teaching can be harmonized with cultural and religious educational values. As Maspul et al. (2025) argued, viewing modernization as a form of cultural stewardship rather than a threat allows educational institutions like pesantren to evolve and thrive[34]. In that spirit, integrating AI can be seen as an extension of the educator’s toolset to better serve the goals of Islamic education – by producing students who are not only knowledgeable in religious texts but also capable of communicating effectively in the Arabic language beyond ritualistic contexts.

The practical implications drawn from this review suggest that with adequate support – in terms of infrastructure, teacher training, and carefully curated content – AI-enhanced Arabic instruction is a feasible and promising endeavor. The recommendations provided here serve as a guide for initial implementation and underscore the importance of monitoring and adapting these innovations to the local context. Stakeholders in Indonesia’s education sector, particularly those overseeing curriculum development in Islamic schools, may consider policies that encourage pilot programs and capacity-building for digital learning tools in language education. International collaboration could also be beneficial, such as sharing best practices from AI in language learning projects across different countries, or co-developing AI tools specifically for less commonly taught languages like Arabic, which cater to diverse learner populations including those in religious schools.

In conclusion, the implementation of artificial intelligence in communicative Arabic language learning holds significant potential to elevate the effectiveness of teaching and learning in pesantren and similar settings. It offers a pathway to resolve longstanding pedagogical challenges by providing immersive, interactive, and personalized learning experiences at scale. By harnessing AI as an ally, educators can better engage the current generation of learners who are digital natives, while firmly grounding language instruction in proven communicative approaches. The outcome we envision is a blended model where traditional strengths of pesantren education – such as strong teacher-student relationships and moral formation – are complemented by the innovative power of AI, resulting in graduates who are confident communicators in Arabic and equipped for the interconnected world. Future research and practice will undoubtedly refine the model, but the evidence so far provides a strong impetus to move forward in this direction. The marriage of AI and CLT in Arabic education is not just a technological upgrade; it is a strategic pedagogical advancement that can contribute to a more effective, enjoyable, and meaningful language learning journey for non-native speakers of Arabic in Indonesia and beyond.

References (Selected)

[1] Azra, A. (2015). Pesantren and globalization: Maintaining tradition amid change. [Conference presentation].

[2] Darancik, Y. (2018). Integrating the four skills: Trends in language teaching. Journal of Language and Education, 4(3), 115–120.

[3] Dörnyei, Z., & Ushioda, E. (2011). Teaching and researching motivation. London: Longman.

[4] Salikhovna, K. K. (2025). The issues of developing students’ linguistic competence in learning Arabic: A literature review. Universal Teaching and Learning Journal, 1(2), 149–160.

[5] Kim, J. (2020). Chatbot-assisted conversational practice: Reducing EFL learners’ speaking anxiety. TESOL Journal, 11(3), e00550.

[6] Yang, W., & Xie, Q. (2022). Improving EFL learners’ speaking skills and willingness to communicate via AI-mediated interactions. System, 108, 102433.

[7] Hapsari, N. D., & Wu, W. C. (2022). Alleviating foreign language speaking anxiety through AI chatbot interactions. Computer Assisted Language Learning. Advance online publication.

[8] Haryanti, T., Oktavia, N. A., & Amir, S. (2024). Arabic language teacher empowerment at Islamic boarding schools through iSpring Suite digital media development training. Journal Ligundi of Community Service, 3(1), 33–41.

[9] Nugraha, A., & Syafe’i, I. (2025). Curriculum foundations for Arabic language education in the AI era: Holistic, juridical, and technological perspectives. Journal of Arabic Language Learning and Teaching, 2(1), 50–66.

[10] Suryaningsih, D. (2019). Embracing digital tools: Modernizing Arabic instruction in Indonesian pesantren. Tadris: Jurnal Pendidikan Bahasa Arab, 11(2), 121–134.

[11] Lukens-Bull, R. A. (2018). Islamic education and modernity in Indonesia. Journal of Indonesian Islam, 12(1), 1–15.

[12] Natheer, S., Elazhary, H., & Al-Ahmadi, H. (2022). Agent-based intelligent tutoring system for Arabic grammar. Journal of King Abdulaziz University: Computing and IT, 11(2), 1–24.

[13] Habash, N. (2010). Introduction to Arabic natural language processing. Morgan & Claypool.

[14] Chen, X., & Lee, S. (2022). AI-driven conversational practice for improving oral fluency and confidence. ReCALL, 34(3), 242–259.

[15] TalkPal. (n.d.). Learn Arabic – TalkPal. Retrieved November 10, 2025, from https://talkpal.ai/learn-arabic

[16] ArabicTutorAI. (n.d.). Fastest Way to Learn Arabic with AI. Retrieved November 10, 2025, from https://www.arabictutorai.com

[17] Aljameel, S. S., O’Shea, J. D., Crockett, K. A., Latham, A., & Kaleem, M. (2017). Development of an Arabic conversational intelligent tutoring system for education of children with ASD. Proceedings of the IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications (CIVEMSA), 24–29.

[18] Dizon, G. (2022). Using AI-powered speech recognition technology to improve English pronunciation and speaking skills. IAFOR Journal of Education: Technology in Education, 10(2), 105–122.

[19] El Shazly, R. (2021). AI-assisted pronunciation training for Arabic EFL learners: A voice assistant approach. IAFOR Journal of Education: Technology in Education, 9(3), 107–126.

[20] Ducar, C., & Schocket, D. (2018). Machine translation and the L2 classroom: Pedagogical solutions for making peace with Google Translate. Foreign Language Annals, 51(4), 779–795.

[21] Qiao, H., & Zhao, A. (2023). Artificial intelligence-based language learning: Illuminating the impact on speaking skills and self-regulation in Chinese EFL context. Frontiers in Psychology, 14, Article 1255594.

[22] Setyono, B., & Wahyuni, A. (2020). The effect of AI-based speaking practice on Indonesian EFL learners’ oral performance and confidence. Journal of English Teaching and Research, 5(2), 123–138.

[23] Horwitz, E. K., Horwitz, M. B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language Journal, 70(2), 125–132.

[24] Zhang, W. (2024). Advances in AI for language learning. Computers & Education, 182, 104545.

[25] Hwang, W.-Y., Luthfi, M. I., & Liu, Y.-F. (2024). AI-enhanced video drama-making for improving writing and speaking skills of students learning English as a foreign language. Innovation in Language Learning and Teaching. Advance online publication. https://doi.org/10.1080/17501229.2024.2437655

[26] Li, B., Tan, Y. L., Wang, C., & Lowell, V. (2025). Two years of innovation: A systematic review of empirical generative AI research in language learning and teaching. Computers & Education: Artificial Intelligence, 2, 100083.

[27] Garba, M. A., & Hassan, A. R. (2024). Use of AI in learning Arabic language by non-Arabic speakers. International Journal of Research and Innovation in Social Science, 8(9), 2519–2532.

[28] Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press.

[29] Lantolf, J. P., & Thorne, S. L. (2006). Sociocultural theory and the genesis of second language development. Oxford University Press.

[30] VanLehn, K. (2011). The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychologist, 46(4), 197–221.

[31] Bloom, B. S. (1984). The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational Researcher, 13(6), 4–16.

[32] Shute, V. J. (2008). Focus on formative feedback. Review of Educational Research, 78(1), 153–189.

[33] Dimitriadou, A., & Lanitis, A. (2023). Smart classrooms and AI: A SWOT analysis. Computer-Assisted Learning Journal, 36(2), 99–113.

[34] Maspul, I., Yusof, N., & Daud, M. (2025). Reimagining Arabic learning in Aceh’s pesantren for a sustainable future. Frontiers in Research, 2(2), 1–12.

[35] Dewaele, J.-M., & Alfawzan, M. (2018). Does foreign language enjoyment outweigh foreign language anxiety in predicting second language performance? Studies in Second Language Learning and Teaching, 8(1), 21–45.