Sage Journals: Discover world-class research

Abstract

This study explores student-initiated translanguaging in the English as a Foreign Language (EFL) context, particularly with AI translation tools, an area largely overlooked in favor of teacher-led approaches. Using a sequential explanatory mixed-method design, the research examined how 69 undergraduates at a Thai university, across various disciplines, integrated AI translation tools into their English learning, focusing on communicative competence within the frameworks of the Optimal Input Hypothesis and Translanguaging Theory. Data were collected through surveys, speaking assessments, and narrative frames, with quantitative data analyzed via path analysis and qualitative data through thematic analysis. The findings reveal a strong reliance on AI tools for comprehension, vocabulary acquisition, and building linguistic confidence, as indicated by high mean scores. Nonetheless, although these tools were effective in enhancing comprehension and shaping student preferences, they also showed a tendency to reduce overall usage due to increased proficiency or potential over-reliance. Qualitative results further accentuated the importance of these tools in decoding complex texts, aiding comprehension, and verifying understanding, with convenience being a major driver of their use. Despite these benefits, path analysis indicated that the impact of AI tools on speaking skills was not statistically significant (β = −.1044, p = .512), raising concerns that frequent reliance on these tools may not substantially improve speaking abilities and might even hinder deeper language mastery. These findings highlight the need for further research to better understand the comprehensive effects of AI-powered translanguaging on language learning outcomes.

Plain language summary

Moving past traditional teaching: How students use AI tools for mixing languages in learning foreign languages

This study looks at how students in Thailand are using AI translation tools to help them learn English, a topic that hasn’t been explored much because most research focuses on traditional, teacher-led methods. The researchers wanted to see how students, rather than teachers, use these AI tools to improve their English, especially in terms of understanding and communicating effectively. They worked with 69 university students from different fields of study and used a combination of surveys, speaking tests, and personal stories to gather information. The results showed that students heavily rely on AI tools to understand English better, learn new words, and feel more confident in their language skills. However, while these tools were great for helping with comprehension, they sometimes led to less overall use as students became more proficient or overly dependent on the technology. The study also found that while AI tools are useful for understanding difficult texts and checking comprehension, they don’t necessarily improve speaking skills. In fact, the analysis suggested that relying too much on these tools might even prevent students from mastering the language fully. Overall, the study suggests that while AI tools have clear benefits, more research is needed to understand their full impact on language learning, especially when it comes to speaking skills.

Keywords

translanguaging AI translation tools input hypothesis communicative competence

Introduction

The integration of artificial intelligence (AI) into language learning, particularly through AI-powered translanguaging, is transforming educational practices by leveraging digital tools to transcend traditional language boundaries (Gašević et al., 2023). Translanguaging, as conceptualized in sociolinguistic theories (Baker, 2011; Wei, 2018), involves the dynamic and flexible use of multiple linguistic resources, fostering deeper learning and critical thinking. Unlike translation, which merely converts text between languages, translanguaging empowers learners to draw from their entire linguistic repertoire to construct meaning and enhance comprehension (Cenoz & Gorter, 2021). Research stresses the pedagogical benefits of teacher-led translanguaging in fostering inclusivity and cognitive engagement in bilingual and multilingual classrooms (Ali, 2024; P. Wang, 2022). Nonetheless, there is a critical gap in understanding the autonomous use of AI tools by students for translanguaging, particularly in English as a Foreign Language (EFL) context where learners interact with digital content independently. Such exploration is essential to evaluate the potential of AI tools in facilitating not just comprehension but also advanced language acquisition, especially within the framework of the Optimal Input Hypothesis, which emphasizes learning through comprehensible yet challenging input (Krashen, 1983, 1989).

Although the body of research on translanguaging in education has largely concentrated on strategies directed by teachers (e.g., Henderson & Ingram, 2018; Heugh et al., 2022; Ramirez & Salinas, 2021), there remains a significant gap concerning student-driven approaches, particularly those utilizing AI translation tools to master complex learning materials. AI-enabled translanguaging tools have the potential to significantly bridge linguistic divides and enhance communicative competence within multilingual learning environments. By enabling fluid transitions between languages, these tools help learners to actively participate in the creation of meaning, thereby nurturing critical thinking and advanced linguistic abilities that surpass the capabilities of traditional machine translation technologies (García & Otheguy, 2020). Recent studies have investigated the role of digital technologies in facilitating translanguaging practices (Lu & Gu, 2024) and the utility of AI tools in supporting independent language learning and comprehension (Kelly & Hou, 2022; E. Y. Kim & Oh, 2023; Kruk & Kałużna, 2025). Despite the recognized value of these tools, their instructional effectiveness is not yet fully understood, especially in terms of promoting learner autonomy and deep engagement with educational content. This study seeks to fill these gaps by critically examining how AI translation tools affect student-led translanguaging, evaluating their impact on developing English communicative skills, and exploring their wider implications for multilingual education settings. The results are intended to inform the ongoing dialog on the integration of AI technologies into language teaching, providing insights that could significantly enhance both the autonomy and linguistic proficiency of learners across diverse educational landscapes.

Literature Review

Conceptual Framework: Translanguaging Theory and Optimal Input Hypothesis

The conceptual framework for this study integrates Translanguaging Theory and the Optimal Input Hypothesis to investigate the role of AI translation tools in supporting language acquisition among Thai students.

Translanguaging Theory

Translanguaging Theory, initially introduced by Williams (1994) and further developed by Baker (2011) and Wei (2018), redefines language use by rejecting the traditional separation of languages in multilingual contexts. Instead, it views multilingual individuals as leveraging their entire linguistic repertoire in a cohesive and dynamic manner to enhance comprehension and learning across modalities such as speaking, reading, and writing (Lewis et al., 2012). Translanguaging is not merely a communication tool but a pedagogical strategy that fosters cognitive engagement, inclusion, creativity, and enhanced problem-solving abilities within educational settings (Cenoz & Gorter, 2020). The theory emphasizes the integration of linguistic, semiotic, and cultural resources to fulfill specific communicative purposes, shifting the perspective of multilingual learners from deficient language users to emerging multilinguals with valuable linguistic assets (García & Kleifgen, 2020; Wei, 2018).

As an instructional approach, translanguaging involves intentional strategies to promote multilingual competence and metalinguistic awareness. These practices range from strong forms, which utilize multiple languages within a single classroom setting, to weaker forms emphasizing cross-linguistic coordination (Cenoz & Gorter, 2022). Research underlines its potential to enhance understanding, creativity, and inclusion in diverse educational contexts, while also highlighting the challenges of developing teachable strategies and assessment methods for effective implementation (Bonacina-Pugh et al., 2021; Canagarajah, 2011; Prilutskaya, 2021). Translanguaging’s adaptability allows educators to integrate it into multilingual classrooms to foster both linguistic and cultural inclusivity, as evidenced by its application in enhancing collaborative and cognitive skills (Cenoz & Gorter, 2020, 2022). Yet, despite its promise, gaps persist in the literature concerning its application with emerging AI technologies, particularly in facilitating autonomous learning and supporting multilingual students in navigating complex linguistic material.

Critiques of translanguaging theory, however, have surfaced in recent literature. MacSwan (2020) and MacSwan and Rolstad (2024) argue that its deconstructivist perspective on language conflicts with civil rights in educational policy and diverges from empirical evidence, potentially undermining pluralist language ideologies. Guerrero (2023) questions the feasibility of integrating translanguaging into two-way immersion programs, citing a lack of longitudinal empirical data. Conversely, scholars such as Wei (2023) and Otheguy et al. (2019) maintain that translanguaging provides a unitary perspective on bilingualism that disrupts dominant language ideologies and fosters linguistic equity. Moreover, Sah and Kubota (2022) and Cinaglia and De Costa (2022) emphasize the need for a critical lens in translanguaging research, addressing power imbalances and aligning with decolonial objectives. Canagarajah (2011) further emphasizes the necessity of exploring translanguaging across domains, particularly with a focus on developing actionable and teachable strategies. Within this contested theoretical landscape, the integration of AI tools into translanguaging practices offers a promising yet underexplored avenue, necessitating further investigation into their pedagogical and practical implications.

Optimal Input Hypothesis

Krashen’s Input Hypothesis, a central component of his Monitor Model, posits that language acquisition is most effective when learners are exposed to comprehensible input slightly beyond their current proficiency levels, a concept known as “i+1” (Krashen, 1989; Scarcella & Perkins, 1987). This theory has profoundly influenced second language acquisition research and pedagogy by emphasizing the role of engaging, linguistically rich, and abundant input in advancing learners’ linguistic competencies, including vocabulary and spelling (Krashen & Mason, 2020). Extensive reading programs, particularly those utilizing graded readers, have demonstrated significant improvements in learners’ writing and overall language skills (Mason, 2019; Tudor & Hafiz, 1989). Moreover, the inclusion of input enhancement techniques, such as Comprehension Aiding Supplementation through drawings and brief translations, has proven effective in improving comprehension and retention, especially when combined with narrative-based approaches like “Story-Listening” (Krashen et al., 2018). The importance of input quality, frequency, and variety has been consistently highlighted across studies, with successful language learners often motivated by the intrinsic pleasure derived from engaging with abundant, meaningful input (Piske & Young-Scholten, 2009; Tsang, 2023). Nevertheless, learners’ strategies frequently adapt to input difficulty, as evidenced by studies on how they manage complex or unfamiliar linguistic material (Bacon, 1992).

Despite its substantial impact on language education and communicative teaching approaches, the Input Hypothesis has faced considerable critique. Critics argue that it lacks a precise operational definition of “comprehensible input” and fails to adequately address phenomena such as fossilization in acquisition-rich environments (Higgs, 1985; Peker & Özkaynak, 2020). Furthermore, some researchers contend that the theory’s simplicity and broad claims limit its applicability in formal classroom settings, where structured teaching and learner output play a critical role (Higgs, 1985; Peker & Arslan, 2020). The hypothesis has also been challenged by alternative theories emphasizing the importance of output and skill-building in language learning (Krashen, 2002; Liu, 2015). Despite these criticisms, recent research continues to validate the significance of comprehensible input in developing specific language skills, particularly listening comprehension, thereby reaffirming the hypothesis’s relevance in modern second language acquisition studies (Kasimo et al., 2024). By fostering a balance between input-driven approaches and the inclusion of strategies that encourage output, language educators can optimize learning experiences and address the limitations identified within Krashen’s framework (Luo, 2024).

Although some educators advocate for monolingual teaching approaches, a growing body of research focuses the pedagogical value of incorporating students’ native languages into the classroom to enhance understanding and engagement (Hall & Cook, 2012; Matiso, 2024; Perales-Fernández-de-Gamboa & Orcasitas-Vicandi, 2025). Elaborated and modified input, which adjusts complexity without oversimplification, has been identified as particularly advantageous for language learning compared to simplified or entirely authentic texts (Long, 2020). The interaction hypothesis further underlines the role of interaction, feedback, and output as essential components of second language acquisition, with output facilitating the noticing of linguistic features and reinforcing learning through active production (Gass & Mackey, 2006; Izumi, 2002). Nevertheless, beginners often rely on form-based processing when initially encountering a new language, necessitating tailored instructional strategies to support their early stages of learning (Apridayani & Waluyo, 2024; Han & Peverly, 2007). As the debate continues over optimal input types and teaching methodologies, researchers increasingly emphasize the importance of integrating diverse approaches to maximize second language acquisition outcomes.

This study investigates how AI translation tools, such as Google Translates (GT), operationalize these theories to enhance language learning for Thai students navigating between Thai and English. According to Translanguaging Theory, these tools amplify comprehension and communication by allowing students to fluidly utilize their complete linguistic toolkit (Kelly & Hou, 2022). In alignment with the Optimal Input Hypothesis, AI translation tools deliver linguistically sophisticated yet accessible input that matches the “i+1” criterion, essential for effective language learning (Krashen, 1989). This bespoke input exposes students to more intricate language structures and vocabulary, enriching their educational experience and fostering superior learning outcomes. These tools are particularly beneficial in higher education EFL contexts, where adapting to individual learning needs is critical (Heugh et al., 2022). Overall, this study explores how AI translation tools not only support translanguaging practices but also fulfill the stringent demands of providing optimal input, thus significantly advancing language acquisition among Thai EFL learners.

AI Translation Tools in Student Foreign Language Learning

The evolution of computer-aided translation (CAT) has been swift and transformative, progressing through distinct phases from germination (1967–1983) to global development (2004–2013), as Sin-wai (2014) delineates. The integration of personal computing in the 1980s catalyzed advancements in multilingual communication, precipitating the development of more sophisticated machine translation (MT) systems by the mid-1990s, including statistical machine translation (SMT; Gaspari, 2024). A significant leap occurred with the advent of neural machine translation (NMT) in the mid-2010s, which employs deep learning to produce translations that are not only more accurate but also contextually nuanced. This innovation has made tools such as Google Translate, DeepL, and Microsoft Translator indispensable in both casual and scholarly communication, with online platforms and cloud-based services further enhancing their efficiency and accessibility at the present time (Schwartz, 2018).

Research into the application of AI translation tools in English language learning has identified a range of benefits and challenges. Klimova et al. (2023) highlight that these tools can substantially improve students’ comprehension and vocabulary acquisition. However, comparative studies indicate that while AI tools and traditional methods both enhance reading and grammar skills, there are no significant differences in the outcomes, suggesting that AI tools’ primary advantage may lie in their convenience and speed rather than educational superiority (H. S. Kim & Cha, 2023; Ting & Tan, 2021). Notwithstanding, students report that these tools help them better understand complex texts and expand their vocabulary more effectively than traditional dictionaries, primarily due to the instantaneity of translations and contextual explanations that minimize cognitive load (Resende & Way, 2021).

Moreover, empirical research by Tsai (2022) in China demonstrated that using Google Translate as a translingual Computer-Assisted Language Learning (CALL) tool in EFL writing enabled students to produce better language and content in revisions than in their original drafts. Relatedly, P. Stapleton and Kin (2019) observed that Google Translate assists primary L2 writers in Hong Kong by generating formal and sophisticated language, thus helping educators demonstrate what constitutes high-quality writing. Conversely, Chung and Ahn (2022) found that the advantages and limitations of Google Translate vary by student proficiency level and text genre, with students expressing high satisfaction and intent to continue using the tool despite its flaws.

Despite the noted advantages, researchers caution against over-reliance on AI translation tools. They argue that while these tools offer significant support for language learning, they also introduce potential risks of dependency and may not always foster a deep understanding of language mechanics (Groves & Mundt, 2015; Murtisari et al., 2024). Thus, while AI translation tools represent a powerful aid in language education, they should be integrated judiciously to complement traditional learning methods rather than replace them.

Translanguaging and Foreign Language Communicative Competence

Translanguaging enhances foreign language communicative competence by allowing learners to use their entire linguistic repertoire dynamically, deepening understanding and improving communication (Lewis et al., 2012). Empirical studies highlight its effectiveness: Kampittayakul (2019) found that 76% of Thai learners improved their interactional competence in English tutorials, demonstrating better communication skills in varied contexts. A quasi-experimental study in Turkey observed significant improvements in students’ speaking skills, attributing gains to the comfort and motivation translanguaging pedagogy provided, allowing students to utilize all linguistic resources effectively (Yasar Yuzlu & Dikilitas, 2022).

Further research indicates that flexible language choice by teachers can encourage student code-switching, enhancing autonomous language use (Adinolfi & Astruc, 2017). In technology-supported environments, students effectively use multimodal resources, aiding intercultural communication and academic success (Ou et al., 2024). However, challenges persist in the broader acceptance of translanguaging. Studies in Lesotho and Turkey reveal that while students benefit from increased confidence and participation, institutional and societal expectations often restrict the pedagogy’s use, not explicitly enhancing communicative competence (Nkhi & Shange, 2024; Yuvayapan, 2019). Similarly, in China, although students prefer multilingual instruction, teachers have mixed feelings about its effectiveness on communicative skills (D. Wang, 2019). These insights suggest that while translanguaging has potential to revolutionize language learning, further investigation is needed to understand its direct impact on communicative competence and its acceptance in educational settings.

The Study

AI translation tools have significantly enhanced student translanguaging practices by facilitating seamless transitions between native languages and English, thereby enriching educational experiences through holistic linguistic engagement (Kelly & Hou, 2022). Recent studies have demonstrated that AI-powered translation tools improve translation accuracy, foster student engagement, and support translanguaging in multilingual educational settings (Kelly & Hou, 2022; Kruk & Kałużna, 2025). These tools empower English as an Additional Language (EAL) learners by enabling them to integrate machine translation into their linguistic repertoires with flexibility and critical awareness, thus enhancing their ability to navigate complex linguistic and content-based challenges (Grieve et al., 2024). In content and language integrated learning contexts, machine translation has been shown to act both as a scaffold to bridge language-content gaps and as an effective tool for language acquisition (E. Y. Kim & Oh, 2023). Nonetheless, current research predominantly emphasizes teacher-led translanguaging (e.g., Henderson & Ingram, 2018; Heugh et al., 2022; Ramirez & Salinas, 2021), with limited exploration of how students autonomously use AI tools to engage with digital content. Moreover, the integration of translanguaging theory and the Optimal Input Hypothesis in English language teaching within higher education remains under-researched. To address these gaps, this study seeks to answer two key questions:

(1) How do students integrate AI translation tools into their foreign language learning practices?

(2) How do AI translation tools impact the development of students’ communication skills?

Method

Research Design, Context, and Participants

This study employed a sequential explanatory mixed-method design (Creswell et al., 2003) to explore the integration of AI translation tools in language learning at a Thai higher education institution. A convenience sampling method (Sedgwick, 2013) was used to select 69 undergraduate students (66.7% female, 33.3% male; mean age = 18.65 years, SD = 0.80) from diverse disciplines, including health, accounting, management, technology, and engineering. These students operated under the university’s English as a Medium of Instruction (EMI) policy, regularly engaging with English texts across courses—88.11% utilized English materials, predominantly in formats such as news and media (55.07%), academic texts (20.29%), and literary works (15.94%). Smartphones emerged as the most commonly used device for accessing translation tools, with 85.51% of participants favoring them for convenience and accessibility. Ethical protocols were rigorously observed, adhering to the Helsinki Declaration, with all participants providing informed written consent after being fully briefed on the study’s objectives. Participation was entirely voluntary and carried no academic repercussions, ensuring the welfare of participants while upholding research integrity.

Teaching Materials and Student Translanguaging Space

The teaching materials for this course, informed by the Optimal Input Hypothesis (Krashen, 1983, 1989), comprised a student e-book containing a collection of fables, as summarized in Figure 1. These stories, selected for their cultural relevance to Thailand and their representation of diverse global traditions, were designed to facilitate the acquisition of both linguistic features of the target language and embedded cultural elements. Over a 9-week period, students engaged in 5 hr of class per week, split into two sessions. The initial session focused on “Story Listening” and story reading to develop receptive skills (listening and reading), while the subsequent session concentrated on applying the target language’s productive skills (writing and speaking; Krashen & Mason, 2020; Krashen et al., 2018; Mason, 2019). Activities included modifying the story plot, summarizing the story, role-playing, group and individual presentations, and analyzing story elements, all aimed at eliciting written and spoken English outputs from students. This pedagogical approach leveraged stories slightly beyond the students’ current proficiency levels (i+1) yet still comprehensible, emphasizing engagement with enriched input (Harmer, 1984).

Figure 1.

Summary of the teaching materials and translanguaging space.

To facilitate student translanguaging (Baker, 2011; Wei, 2018; Williams, 1994), —navigating between Thai and English—smart gadgets, for example, smartphones and tablets, were extensively utilized throughout the learning process, through which students accessed the AI translation tools following their preferences, such as Google Translate and other applications. Each week, students encountered multiple stories, with additional supplementary tales accessible via QR codes embedded in the course e-book. This feature enabled instructors to customize content to meet varying student needs, ensuring that more advanced students could access further challenging material. The selected primary fables, such as “The Lark and Her Young Ones,”“The Fox Without a Tail,”“The Boy and the Apple Tree,” and “The Most Beautiful Garden,” were chosen for their narrative depth and moral insights, which support language learning alongside critical thinking and ethical reflection. The e-book compiled all learning materials, including stories from various sources like Aesop’s Fables websites, YouTube, and other educational platforms, available in text, audio, and video formats to accommodate different learning preferences. This multimedia approach not only enriched the learning experience but also facilitated active student engagement and translanguaging practices during both receptive and productive learning activities.

Students’ translanguaging behavior encompasses the dynamic and flexible use of their full linguistic repertoire to facilitate meaning-making, enhance learning outcomes, and foster engagement across languages (Mendoza, 2022; Palfreyman & Al-Bataineh, 2018). Research demonstrates that this practice supports not only content comprehension but also critical thinking, collaborative problem-solving, and peer interaction, thereby fostering a more inclusive and effective learning environment (Ali, 2024; Hernandez Garcia et al., 2023; Mendoza, 2022). Evidence from classroom interactions, particularly in this study, highlights how students seamlessly transition between listening, reading, and preparing for productive activities, such as storytelling or discussions, utilizing translanguaging as a means to navigate complex tasks and deepen their understanding of content (Cenoz & Gorter, 2020; Lewis et al., 2012). Such practices stress the potential of translanguaging to integrate linguistic diversity into educational settings, offering a strong framework for promoting active participation and collaborative learning among students. Figure 2 is one translanguaging activity among the students.

Figure 2.

One sample of the students’ translanguaging practices on their tablets.

Data Collection: Instruments, Measures, Reliability, and Validity

At the end of the course, the researcher used Google Forms to distribute a questionnaire to assess students’ translanguaging behaviors with AI translation tools. The questionnaire, based on insights from recent studies (Chung & Ahn, 2022; H. S. Kim & Cha, 2023; Klimova et al., 2023; P. Stapleton & Kin, 2019; Ting & Tan, 2021; Tsai, 2022), consisted of 10 items translated fully into Thai for clarity. As presented in Figure 3, examples of items included “I use digital translation tools to help me understand English texts” and “Digital translation tools are effective in helping me understand the meaning of difficult English words.” Responses were collected on a Likert scale from 1 (Strongly Disagree) to 5 (Strongly Agree), allowing detailed analysis of student preferences and behaviors.

Figure 3.

Descriptive statistics.

The reliability of the survey was high, with Cronbach’s alpha values reaching .908, well above the .70 threshold indicative of high internal consistency. The survey’s validity was affirmed through face validity checked by an expert in English Language Teaching and Exploratory Factor Analysis (EFA) following C. D. Stapleton’s (1997) guidelines. The KMO and Bartlett’s tests resulted in a chi-square of 453.105 (p < .001) and a sampling adequacy of 0.869, both significantly surpassing the .50 benchmark, confirming the survey’s structural validity and reliability. These measures ensure the methodological soundness of the research instrument used in the study.

The survey also gathered data on the use of AI translation tools among students, with results revealing high engagement levels with English texts and significant long-term integration of these tools into their study routines. Specifically, 49.28% of students reported using translation tools for “5 to 6 years or more,” while 42.03% have been using them for “3 to 4 years,” indicating that the majority have well-established usage habits. Regarding frequency, 52.17% of participants used these tools “Often” and 39.13% “Sometimes,” highlighting a strong reliance on these tools for comprehending English texts. This data was further analyzed using inferential statistics, as detailed in the results section, illustrating that AI translation tools are not just occasional aids but integral to the students’ daily educational activities.

To evaluate communicative competence, students participated in a speaking assessment where they retold English stories previously studied in class. They utilized Canva.com to present story-related visuals, enhancing the narrative delivery. This assessment, conducted within a typical classroom setting, lasted between 10 and 15 min per student. An assessment rubric was employed to evaluate fluency, pronunciation, vocabulary, and grammar, with each component graded on a scale from 0 (minimum) to 2.5 (maximum), leading to a maximum possible score of 10. This structured approach allowed for a comprehensive evaluation of each student’s speaking proficiency in a controlled academic environment.

Furthermore, narrative frames, crucial to the narrative inquiry approach (Barkhuizen, 2014), provided students with a structured format to articulate their experiences using Thai to understand English texts and leveraging digital technology for translation. Students were prompted to write reflective compositions in either Thai or English, incorporating personal anecdotes to deepen the narrative context. This option to use dual languages likely enriched the authenticity and depth of their responses, thus enhancing the qualitative aspect of the data. This qualitative method complemented the quantitative data derived from Likert-scale surveys, offering nuanced insights into the complex interplay between language learning and technology use. The integration of personal narratives with quantitative analysis provided a holistic view of how translanguaging and digital tools affect language comprehension, bridging the gap between statistical data and individual learner experiences.

Data Analysis

The survey data were analyzed using descriptive and inferential statistical methods to provide a comprehensive overview and detailed insights. Descriptive statistics, including means and standard deviations, summarized central tendencies and variability, while inferential techniques such as Exploratory Factor Analysis (EFA) and Path Analysis probed the underlying structures and relationships among variables, including scores from speaking assessments. Concurrently, the qualitative data were processed through thematic analysis, following the framework set by Clarke and Braun (2017), to identify key themes and patterns that elucidated participants’ experiences and perceptions. This integrated approach ensured a thorough analysis, linking quantitative metrics with qualitative insights to enrich the study’s findings.

Results

Descriptive Statistics

The descriptive statistics for the integration of AI translation tools in student translanguaging behaviors indicate a strong reliance on these digital aids, with an overall high mean score of 4.07 (SD = 0.67), as shown in Figure 2. This suggests that students heavily utilize digital translation tools, which significantly enhance their language learning and understanding. Notably, students demonstrate a preference for digital translation tools over traditional paper dictionaries, with a mean score of 4.38 (SD = 0.89), and they find these tools particularly effective for learning new vocabulary (M = 4.35, SD = 0.84).

Further analysis reveals that students use digital tools extensively for various academic purposes: checking their understanding of English passages scores a mean of 4.20 (SD = 0.83), general understanding of English texts at 4.16 (SD = 0.78) and increasing confidence in reading English at 4.14 (SD = 0.88). Additionally, these tools are instrumental in helping students grasp difficult English words, with a mean score of 4.12 (SD = 0.87). There is also a noted improvement in understanding English texts when using digital translation tools (M = 4.03, SD = 0.89), and students frequently use these tools to summarize English texts into Thai (M = 3.96, SD = 0.95).

Regular usage of digital dictionaries or translation apps is reported with a mean score of 3.59 (SD = 1.16), while reliance on translation apps for converting sentences or paragraphs from English to Thai is also significant (M = 3.81, SD = 0.94). These statistics reflect the critical role digital translation tools play in enhancing students’ comprehension, confidence, and vocabulary acquisition, pointing to their extensive integration into the students’ academic routines and highlighting their effectiveness in supporting advanced translanguaging practices.

Inferential Statistics

The Exploratory Factor Analysis (EFA) conducted on the integration of AI translation tools in student translanguaging behavior revealed two distinct factors. This analysis, validated by a Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy of 0.869, indicated excellent sampling adequacy. Furthermore, Bartlett’s Test of Sphericity supported this adequacy with a highly significant result (Chi-Square = 453.105, p < .001). The Total Variance Explained table highlighted initial eigenvalues of 5.748 and 1.157 for these factors, which account for 57.48% and 11.57% of the variance, respectively, totaling 69.06%. After factor extraction, these percentages adjusted to 54.10% and 6.79%, cumulatively 60.89%. Subsequent rotation refined the factor structure, identifying “Use of Digital Translation Tools for Comprehension” and “Preference for Digital Translation Tools” as primary factors, explaining 36.10% and 24.78% of the variance, respectively. This delineation stresses the integral roles of digital tools in enhancing comprehension and shaping preferences in translanguaging practices.

A subsequent path analysis aimed to delineate the relationships among key variables concerning the use of digital translation tools, including Frequency of Using Translation Tools, Use of Digital Translation Tools for Comprehension (UDTC), Preference for Digital Translation Tools (FDTT), and Digital Technology Functions in Student Translanguaging Behavior (DTF). The analysis demonstrated significant positive direct paths from DTF to both UDTC (coefficient of 0.99, p < .001) and FDTT (coefficient of 0.75, p < .001), indicating that enhanced engagement with digital functions markedly increases both comprehension use and tool preference. Conversely, a significant negative relationship was found between UDTC and Frequency of Using Translation Tools (coefficient of −2.74, p = .026), suggesting that more intensive use of digital tools for comprehension might reduce the overall frequency of tool usage, potentially reflecting either an over-reliance or a decreased necessity for frequent tool use.

However, the link between FDTT and the Frequency of Using Translation Tools was not significant (coefficient of −0.26, p = .295), indicating that preference for digital tools does not necessarily affect how often they are used. Additionally, a direct positive path from DTF to the Frequency of Using Translation Tools (coefficient of 2.47, p = .070) implies that frequent engagement with digital functions could lead to increased tool usage, although this finding was only marginally significant. The model’s goodness of fit was confirmed with a Root Mean Square Error of Approximation (RMSEA) of 0.0664, affirming the structural validity of the explored relationships. Table 1 and Figure 4 below provide the summary and illustration of the path analysis results.

Table 1.

Summary of the Path Analysis Results.

Path	Coefficient (β)	p-Value	Significance
DTF → UDTC	.99	<.001	Strongly significant
DTF → FDTT	.75	<.001	Strongly significant
UDTC → Frequency of Using Translation Tools	−2.74	.026	Significant
FDTT → Frequency of Using Translation Tools	−.26	.295	Not significant
DTF → Frequency of Using Translation Tools	2.47	.07	Marginally significant

Figure 4.

Path diagram for AI tool functions in student translanguaging behavior.

Thematic Analysis

As presented in Table 2, thematic analysis of student narratives from digital storytelling sessions identified five major themes demonstrating their reliance on AI translation tools for English learning. First, Vocabulary Translation is a prevalent theme where students use digital translation tools to decipher unfamiliar or complex words within texts. This method not only facilitates the understanding of specific terms but also enhances vocabulary acquisition, as students often use these tools to swiftly uncover the meanings of new or challenging words, thereby improving their overall comprehension. Another significant theme is Sentence Translation, where some students resort to translating entire sentences or paragraphs to better understand the structure and meaning of complex English constructs. This strategy enables them to comprehend the full context and flow of the text, proving particularly beneficial when individual word translations are insufficient for full comprehension. These findings accentuate the critical role of digital translation tools in supporting students’ language learning by providing immediate assistance with vocabulary and sentence structure, thus aiding in the deeper understanding of English texts.

Use it to look up new vocabulary I don’t know. (S5)

When I see a word I don’t understand, I use a translation tool to get the meaning. (S10)

Use Google Translate to translate the entire sentence. (S6)

When a paragraph is too complex, I translate it to see the full meaning. (S12)

Table 2.

Summary of the Emergent Themes.

Theme	Keywords	Frequency	Sample students’ statements
Vocabulary Translation	Translate, unfamiliar, words	12	“Translate unfamiliar words.” (Student 4)
Sentence Translation	Translate, sentences, paragraphs	8	“Translate text that I don’t understand.” (Student 3)
Comprehension Aid	Understand, meaning, easier	10	“Use it to understand the meaning of sentences.” (Student 8)
Verification	Compare, check, accuracy	6	“Read and compare sentences.” (Student 2)
Convenience	Quick, easy, accessible	5	“Use Google Translate for quick translations.” (Student 6)

Thematic analysis also highlighted two additional themes related to students’ use of digital technology in language learning: Comprehension Aid and Verification. Many students employ digital translation tools such as Comprehension Aids to grasp the overall meaning of sentences or paragraphs. This strategy involves using translations to either clarify ambiguities or confirm their interpretations of the text, effectively enhancing their understanding of complex English passages. Digital tools serve as crucial aids in this process, allowing students to secure a clearer and more accurate comprehension, which supports their overall learning trajectory. The theme of Verification further demonstrates students’ strategic use of digital tools to confirm the accuracy of their interpretations. By comparing their own translations with those generated by digital tools, students can validate their understanding, correcting any discrepancies. This practice not only bolsters their confidence but also ensures a deeper and more reliable comprehension of complex texts, facilitating a more engaged and effective learning experience. These themes collectively accentuate the integral role of digital translation tools in enhancing linguistic comprehension and the verification of learning among students.

It helps me make sense of difficult sentences. (S13)

I use digital tools to ensure I understand the paragraph correctly. (S17)

Check my translation with the tool to ensure it’s right. (S9)

I use translation tools to verify my understanding. (S14)

The theme of Convenience emerged prominently in students’ narratives, highlighting their appreciation for the quick and easy access provided by digital translation tools. This convenience is a critical factor for students, as it significantly enhances the efficiency of translating and understanding complex texts. The ability to obtain immediate translations not only streamlines the learning process but also reduces the time and effort required to grasp challenging material, making these tools indispensable in students’ academic routines. This emphasis on convenience emphasizes the practical benefits of digital translation tools in facilitating more efficient and accessible language learning.

It’s easy to use and helps me understand quickly. (S11)

I like the convenience of having translations at my fingertips. (S15)

Figure 5 visually represents how these themes interrelate and support each other in the process of understanding complex English texts using digital translation tools.

Vocabulary Translation and Sentence Translation are directly linked to Comprehension Aid, indicating that translating specific words or entire sentences helps students understand the overall meaning of complex texts.

Comprehension Aid is connected to Verification, showing that once students have an initial understanding, they use digital tools to verify their interpretations.

Verification is linked to Convenience, highlighting that the ease and accessibility of digital tools play a role in confirming understanding.

Convenience is also connected back to Comprehension Aid, demonstrating that the quick and easy access to translations helps in comprehending complex texts efficiently.

Figure 5.

Relationships among the emergent themes.

Impacts on the Productive Skill

A path analysis was conducted to examine the relationships among Frequency of Reading English Texts, Student Translanguaging Behavior in Digital Storytelling Instructions (STB), Frequency of Using Translation Tools, and Digital Technology Functions in Student Translanguaging Behavior (DTF) to understand their impact on students’ Speaking Scores in digital storytelling. The analysis revealed that Frequency of Reading English Texts had a slight positive effect on Speaking Score (coefficient = 0.0607), though this effect was not statistically significant (p = .517), suggesting that reading frequency may have a limited influence on speaking performance in this context. Meanwhile, Student Translanguaging Behavior (STB) showed a negative coefficient of −0.2338 (p = .132), implying a potential negative impact of increased translanguaging on speaking performance; however, this relationship was also not statistically significant. These findings suggest that while translanguaging is prevalent, its effectiveness in improving speaking skills remains unclear and warrants further investigation.

The analysis revealed that the Frequency of Using Translation Tools had a negative impact on Speaking Score, with a coefficient of −0.1044 (p = .512), indicating that frequent use of these tools does not necessarily enhance speaking skills. This non-significant result may suggest that reliance on translation tools could hinder language mastery, which is essential for effective speaking. In contrast, Digital Technology Functions in Student Translanguaging Behavior (DTF) showed a positive coefficient of 0.2144 (p = .247), suggesting a potential beneficial effect of digital technology use in translanguaging, though this relationship was also not statistically significant. The overall model fit, as assessed by the Root Mean Square Error of Approximation (RMSEA), was 0.0867, slightly exceeding the acceptable threshold of 0.08, indicating that the model may not fully capture the complexities of the relationships among the variables. This suggests that further refinement of the model or the inclusion of additional variables may be necessary to better understand the factors influencing speaking performance in the context of digital storytelling. Table 3 and Figure 6 below provide the summary and path diagram to the speaking performance.

Table 3.

Summary of the Path Analysis on the Speaking Performance.

Path	Coefficient (β)	p-Value	Significance	Explanation
Frequency of Reading English Texts → Speaking Score	.0607	.517	Not significant	Slight positive effect, not statistically significant.
STB → Speaking Score	−.2338	.132	Not significant	Negative effect, not statistically significant.
Frequency of Using Translation Tools → Speaking Score	−.1044	.512	Not significant	Negative effect, not statistically significant.
DTF → Speaking Score	.2144	.247	Not significant	Positive effect, not statistically significant.

Figure 6.

Path diagram to the speaking performance.

Discussion and Implications

AI Translation Tools in Student Translanguaging Behaviors

The integration of AI translation tools into students’ foreign language learning practices, as demonstrated by the findings of this study, highlights their transformative role in enhancing comprehension, vocabulary acquisition, and overall linguistic confidence. The high mean scores across various metrics, including the preference for digital translation tools over traditional paper dictionaries (M = 4.38, SD = 0.89), reflect a significant shift in how students approach language learning. This shift is supported by Translanguaging Theory, which posits that learners benefit from leveraging their entire linguistic repertoire, facilitated by tools that provide immediate and contextually nuanced translations (Cenoz & Gorter, 2020; Waluyo & Kusumastuti, 2024; Wei, 2018). Moreover, the high level of reliance on these tools for checking comprehension and understanding English texts (M = 4.20, SD = 0.83) aligns with the Optimal Input Hypothesis, which emphasizes the importance of exposure to input that is both comprehensible and slightly challenging (Krashen, 1983; Krashen & Mason, 2020). The effectiveness of these tools in providing such input is further corroborated by recent advancements in neural machine translation (NMT) technologies, which have made AI translation tools, for example, Google Translate, more accurate and accessible, thereby facilitating their widespread adoption in academic settings (Gaspari, 2024; Schwartz, 2018).

The thematic analysis of student narratives further reinforces the quantitative findings by illustrating the strategic use of AI translation tools in enhancing students’ comprehension and language production. The themes of Vocabulary Translation and Sentence Translation reveal that students frequently use these tools to decode complex words and sentence structures, enabling them to grasp the full context of English texts. This strategic use aligns with Mason’s (2019) application of the Optimal Input Hypothesis in EFL contexts, where similar tools have been shown to bridge proficiency gaps by providing students with access to challenging yet comprehensible input. Furthermore, the themes of Comprehension Aid and Verification highlight the critical role of these tools in confirming students’ interpretations and clarifying ambiguities, thereby supporting more engaged and effective learning experiences (P. Stapleton & Kin, 2019; Tsai, 2022; Zaim et al., 2024). The emphasis on convenience, as noted by students, underlines the practical benefits of AI translation tools, particularly their ability to streamline the learning process and reduce cognitive load, a point echoed in the broader literature on the evolution of translation technologies (Resende & Way, 2021; Sin-wai, 2014).

Nonetheless, the findings also point to potential risks associated with the over-reliance on AI translation tools, particularly concerning the development of deeper linguistic processing skills. The negative relationship between the Use of Digital Translation Tools for Comprehension (UDTC) and the Frequency of Using Translation Tools suggests that intensive use of these tools might lead to decreased overall usage, potentially reflecting either growing proficiency or a dependency that reduces the need for frequent engagement (Groves & Mundt, 2015). This aligns with concerns raised by Murtisari et al. (2024), who caution against the uncritical integration of AI translation tools in language education. While these tools offer significant benefits, their overuse could undermine students’ ability to engage critically with language, thereby limiting their development as autonomous language learners. Thus, the discussion emphasizes the need for a balanced approach that integrates AI translation tools judiciously within a broader pedagogical framework, ensuring that they complement rather than replace traditional learning methods. This approach aligns with the principles of Translanguaging Theory and the Optimal Input Hypothesis, both of which advocate for a holistic and dynamic approach to language learning that leverages all available linguistic and technological resources (Baker, 2011; Kelly & Hou, 2022).

Impacts of Using AI Translation Tools Students’ Communication Skills

The path analysis reveals a complex relationship between the frequency of using AI translation tools and students’ speaking performance. Specifically, the negative coefficient associated with the Frequency of Using Translation Tools suggests that frequent reliance on these tools may not contribute positively to speaking skills. This result aligns with existing literature that cautions against over-reliance on AI translation tools, as they may impede the development of deeper language mastery necessary for effective verbal communication (Groves & Mundt, 2015). Despite the potential of AI tools to enhance comprehension and vocabulary, their use does not appear to directly translate into improved productive skills, such as speaking, which are critical for communicative competence.

Moreover, the non-significant negative impact of Student Translanguaging Behavior (STB) on Speaking Scores highlights the complex role that translanguaging plays in communication skill development. While translanguaging has been shown to enhance learners’ interactional competence and flexibility in language use (Lewis et al., 2012; Yasar Yuzlu & Dikilitas, 2022), its effectiveness in improving speaking performance within this study’s digital storytelling context remains unclear. This ambiguity may stem from the dual nature of translanguaging, which, while promoting a deeper understanding and comfort in language use, may also lead to over-reliance on one’s first language, potentially limiting the practice and fluency needed for effective communication in the target language (D. Wang, 2019). This finding echoes the challenges identified by Nkhi and Shange (2024), where institutional and societal expectations restrict the full benefits of translanguaging, particularly in enhancing communicative competence.

Conversely, the positive yet non-significant coefficient for Digital Technology Functions in Student Translanguaging Behavior (DTF) suggests that digital tools, when integrated thoughtfully into translanguaging practices, could potentially support the development of communication skills. This finding is in line with research by Ou et al. (2024), which emphasizes the role of technology-supported environments in leveraging multimodal resources to aid intercultural communication and academic success. However, the lack of statistical significance in this study indicates that while digital tools may hold promise, their impact on speaking skills is not straightforward and may depend on how these tools are utilized within the pedagogical framework. The overall model’s RMSEA value of 0.0867 further suggests that the current model may not fully capture the complexities involved in these relationships, indicating a need for further research and model refinement. This complexity highlights the ongoing debate in literature regarding the efficacy of translanguaging and digital tools in enhancing communicative competence, suggesting that their role in language education requires careful consideration and further empirical exploration (Adinolfi & Astruc, 2017; Kampittayakul, 2019).

Conclusion, Limitation, and Recommendation

This study explored the integration of AI translation tools into student translanguaging practices and their impact on language acquisition, particularly in enhancing comprehension and communicative skills. The findings demonstrate that AI translation tools play a critical role in supporting students’ engagement with complex linguistic material, improving vocabulary acquisition, comprehension, and overall confidence in reading English. Students frequently relied on these tools for translating words, sentences, and paragraphs, leveraging their functionality to aid in comprehension, verify interpretations, and streamline the learning process. Factor analysis highlighted two primary dimensions: the use of digital tools for comprehension and students’ preferences for these tools, emphasizing their significance in academic routines. However, the study revealed ambiguous effects on productive skills, such as speaking, where frequent reliance on translation tools showed no significant positive correlation with speaking performance. These findings highlight the complexity of translanguaging practices mediated by AI tools, necessitating further research to understand their broader implications.

Despite its contributions, the study has limitations that warrant attention. Methodologically, the non-significant effects of AI tools on speaking performance suggest potential challenges in using translation tools for oral skill development. Students may rely excessively on digital aids for written comprehension, which might inhibit spontaneous language production and fluency in speech. The ambiguous relationship between translanguaging and speaking skills underlines the need for longitudinal studies to assess the long-term impacts of AI tools on productive abilities. Additionally, the study’s reliance on self-reported data and a narrow focus on higher education students in a specific cultural context may limit the generalizability of its findings. Future research should adopt mixed-method approaches, combining quantitative and qualitative data, to capture a more nuanced understanding of how AI tools influence speaking skills. Expanding the sample to include diverse linguistic and educational contexts could also provide more comprehensive insights into the interplay between translanguaging and productive language skills.

To optimize the pedagogical integration of AI translation tools, refinements in analytical models and instructional strategies are essential. Analytical models could be enhanced by incorporating variables such as cognitive load, interactional opportunities, and individual learner differences to better capture the dynamics of translanguaging and speaking performance. Pedagogical strategies should emphasize balanced use, encouraging students to gradually reduce dependence on translation tools while fostering autonomous language production through activities such as role-playing, debate, and spontaneous speaking tasks. Educators can also design scaffolded interventions that integrate AI tools into task-based learning, where students alternate between guided use of translation tools and independent communicative practice. By refining both the theoretical frameworks and instructional applications, future studies and teaching practices can unlock the full potential of AI tools to enhance language learning in diverse educational contexts.

Footnotes

ORCID iDs

Budi Waluyo

Fouzia Rouaghe

Ethical Considerations

This study was conducted in accordance with the ethical standards set forth by Walailak University and adhered to the principles of the Declaration of Helsinki.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Data Availability Statement

Data sharing not applicable to article as no datasets were generated or analyzed during the current study.

References

Adinolfi

Astruc

(2017). An exploratory study of translanguaging practices in an online beginner-level foreign language classroom. Language Learning in Higher Education, 7(1), 185–204.

Ali

A. D.

(2024). Translanguaging in multilingual university classrooms: Effects on students’ language skills and perceptions. Bilingual Research Journal, 47(2), 186–210. https://doi.org/10.1080/15235882.2024.2336942

Apridayani

Waluyo

(2024). Antecedents and effects of students’ enjoyment and boredom in synchronous online English courses. Journal of Multilingual and Multicultural Development, 45(10), 4254–4269. https://doi.org/10.1080/01434632.2022.2152457

Bacon

S. M.

(1992). Authentic listening in Spanish: How learners adjust their strategies to the difficulty of the input. Hispania, 75(2), 398–412. https://doi.org/10.2307/344077

Baker

(2011). Foundations of bilingual education and bilingualism (5th ed.). Multilingual Matters.

Barkhuizen

(2014). Narrative research in language teaching and learning. Language Teaching, 47(4), 450–466.

Bonacina-Pugh

da Costa Cabral

Huang

(2021). Translanguaging in education. Language Teaching, 54(4), 439–471. https://doi.org/10.1017/S0261444821000173

Canagarajah

(2011). Translanguaging in the classroom: Emerging issues for research and pedagogy. Applied Linguistics Review, 2, 1–28. https://doi.org/10.1515/9783110239331.1

Cenoz

Gorter

(2020). Pedagogical translanguaging: An introduction. System, 92, Article 102269.

10.

Cenoz

Gorter

(2021). Pedagogical translanguaging. Cambridge University Press.

11.

Cenoz

Gorter

(2022). Pedagogical translanguaging and its application to language classes. RELC Journal, 53(2), 342–354. https://doi.org/10.1177/00336882221082751

12.

Chung

E. S.

Ahn

(2022). The effect of using machine translation on linguistic features in L2 writing across proficiency levels and text genres. Computer Assisted Language Learning, 35(9), 2239–2264.

13.

Cinaglia

De Costa

P. I.

(2022). Cultivating critical translingual awareness: Challenges and possibilities for teachers and teacher educators. RELC Journal, 53(2), 452–459. https://doi.org/10.1177/00336882221113659

14.

Clarke

Braun

(2017). Thematic analysis. The Journal of Positive Psychology, 12(3), 297–298.

15.

Creswell

J. W.

Clark

V. L. P.

Gutmann

M. L.

Hanson

W. E.

(2003). Advanced mixed. Handbook of Mixed Methods in Social & Behavioral Research, 209, 209–240.

16.

García

Kleifgen

J. A.

(2020). Translanguaging and literacies. Reading Research Quarterly, 55(4), 553–571. https://doi.org/10.1002/rrq.286

17.

García

Otheguy

(2020). Plurilingualism and translanguaging: Commonalities and divergences. International Journal of Bilingual Education and Bilingualism, 23(1), 17–35. https://doi.org/10.1080/13670050.2019.1598932

18.

Gašević

Siemens

Sadiq

(2023). Empowering learners for the age of artificial intelligence. Computers and Education: Artificial Intelligence, 4, Article 100130.

19.

Gaspari

(2024). The history of translation technologies. In Lange

Monticelli

Rundle

(Eds.), The Routledge handbook of the history of translation studies (pp. 324–338). Routledge.

20.

Gass

S. M.

Mackey

(2006). Input, interaction and output: An overview. AILA Review, 19(1), 3–17. https://doi.org/10.1075/aila.19.03gas

21.

Grieve

Rouhshad

Petraki

Bechaz

Dai

D. W.

(2024). Nursing and midwifery students’ ethical views on the acceptability of using AI machine translation software to write university assignments: A deficit-oriented or translanguaging perspective? Journal of English for Academic Purposes, 70, Article 101379. https://doi.org/10.1016/j.jeap.2024.101379

22.

Groves

Mundt

(2015). Friend or foe? Google Translate in language for academic purposes. English for Specific Purposes, 37, 112–121.

23.

Guerrero

(2023). Gauging the adequacy of translanguaging allocation policy in two-way immersion programs in the US. Journal of Latinos and Education, 22(4), 1427–1441. https://doi.org/10.1080/15348431.2021.1971086

24.

Hall

Cook

(2012). Own-language use in language teaching and learning. Language Teaching, 45(3), 271–308. https://doi.org/10.1017/S0261444812000067

25.

Han

Peverly

S. T.

(2007). Input processing: A study of Ab Initio learners with multilingual backgrounds. International Journal of Multilingualism, 4(1), 17–37. https://doi.org/10.2167/ijm052.0

26.

Harmer

(1984). Krashen’s Input Hypothesis and the teaching of EFL. World Englishes, 3(1), 11–15.

27.

Henderson

K. I.

Ingram

(2018). “Mister, you’re writing in Spanglish”: Fostering spaces for meaning making and metalinguistic connections through teacher translanguaging shifts in the bilingual classroom. Bilingual Research Journal, 41(3), 253–271. https://doi.org/10.1080/15235882.2018.1481894

28.

Hernandez Garcia

Schleppegrell

M. J.

Sobh

Monte-Sano

. (2023). The translanguaging school. Phi Delta Kappan, 105(2), 8–12. https://doi.org/10.1177/00317217231205934

29.

Heugh

French

Arya

Pham

Tudini

Billinghurst

Tippett

Chang

L. C.

Nichols

Viljoen

J. M.

(2022). Multilingualism, translanguaging and transknowledging: Translation technology in EMI higher education. AILA Review, 35(1), 89–127.

30.

Higgs

T. V.

(1985). The input hypothesis: An inside look. Foreign Language Annals, 18(3), 197–203. https://doi.org/10.1111/j.1944-9720.1985.tb01791.x

31.

Izumi

(2002). Output, input enhancement, and the noticing hypothesis: An experimental study on ESL relativization. Studies in Second Language Acquisition, 24(4), 541–577. https://doi.org/10.1017/S0272263102004023

32.

Kampittayakul

(2019). The role of translanguaging in improving Thai learners’ interactional competence in dyadic “English as a foreign language” tutorial sessions. PASAA: Journal of Language Teaching and Learning in Thailand, 56, 80–111.

33.

Kasimo

Y. Y.

Ota

M. K.

Kara

Y. M. K.

(2024). Input hypotheses for developing EFL/ESL learners’ listening comprehension skills. Journal of Research in Instructional, 4(2), 367–379. https://doi.org/10.30862/jri.v4i2.436

34.

Kelly

Hou

(2022). Empowering learners of English as an additional language: Translanguaging with machine translation. Language and Education, 36(6), 544–559. https://doi.org/10.1080/09500782.2021.1958834

35.

Kim

E. Y.

(2023). Machine translation use as translanguaging in content and language integrated learning: A case study in a General English Course for global citizenship. English Teaching, 78(4), 59–82. https://doi.org/10.15858/engtea.78.4.202312.59

36.

Kim

H. S.

Cha

(2023). Integrating Al translators into L2 Reading Classes. Multimedia-Assisted Language Learning, 26(3), 9.

37.

Klimova

Pikhart

Benites

A. D.

Lehr

Sanchez-Stockhammer

(2023). Neural machine translation in foreign language teaching and learning: A systematic review. Education and Information Technologies, 28(1), 663–682.

38.

Krashen

(1983). Second language acquisition theory and the preparation of teachers: Toward a rationale. In Alatis

J. E.

Stern

H. H.

Strevens

(Eds.), Georgetown university round table on languages and linguistics (pp. 255–263). Georgetown University Press.

39.

Krashen

(1989). We acquire vocabulary and spelling by reading: Additional evidence for the input hypothesis. The Modern Language Journal, 73(4), 440–464.

40.

Krashen

(2002). The comprehension hypothesis and its rivals. In English Teachers Association/ROC (Ed.), Selected papers from the eleventh international symposium on English teaching/fourth Pan-Asian conference (pp. 395–404). Crane Publishing Company.

41.

Krashen

Mason

(2020). The optimal input hypothesis: Not all comprehensible input is of equal value. CATESOL Newsletter, 5(1), 1–2.

42.

Krashen

Mason

Smith

(2018). Some new terminology: Comprehensionaiding supplementation and form-focusing supplementation. Language Learning and Teaching, 60(6), 12–13.

43.

Kruk

Kałużna

(2025). Investigating the role of AI tools in enhancing translation skills, emotional experiences, and motivation in L2 learning. European Journal of Education, 60(1), Article e12859. https://doi.org/10.1111/ejed.12859

44.

Lewis

Jones

Baker

(2012). Translanguaging: Developing its conceptualisation and contextualisation. Educational Research and Evaluation, 18(7), 655–670.

45.

Liu

(2015). A critical review of Krashen’s input hypothesis: Three major arguments. Journal of Education and Human Development, 4(4), 139–146. https://doi.org/10.15640/jehd.v4n4a16

46.

Long

M. H.

(2020). Optimal input for language learning: Genuine, simplified, elaborated, or modified elaborated? Language Teaching, 53(2), 169–182. https://doi.org/10.1017/S0261444819000466

47.

M. M.

(2024). Review of research on digital translanguaging among teachers and students: A visual analysis through CiteSpace. System, 123, 1–14. https://doi.org/10.1016/j.system.2024.103314

48.

Luo

(2024). A review of Krashen’s input theory. Journal of Education, Humanities and Social Sciences, 26, 130–135. https://pdfs.semanticscholar.org/fdf1/a848544579ce384acaecd572c243aad1bf03.pdf

49.

MacSwan

(2020). Translanguaging, language ontology, and civil rights. World Englishes, 39(2), 321–333. https://doi.org/10.1111/weng.12464

50.

MacSwan

Rolstad

(2024). (Un) grounded language ideologies: A brief history of translanguaging theory. International Journal of Bilingualism, 28(4), 719–743. https://doi.org/10.1177/13670069241236703

51.

Mason

(2019). Guided SSR before self-selected reading. Shitennoji University Bulletin, 67, 445–456.

52.

Matiso

N. H.

(2024). Optimising culturally responsive pedagogies in multicultural English second language classrooms. International Journal of Learning, Teaching and Educational Research, 23(11), 384–401. https://doi.org/10.26803/ijlter.23.11.20

53.

Mendoza

(2022). What does translanguaging-for-equity really involve? An interactional analysis of a 9th grade English class. Applied Linguistics Review, 13(6), 1055–1075. https://doi.org/10.1515/applirev-2019-0106

54.

Murtisari

E. T.

Kristianto

A. K.

Bonar

(2024). Self-directed use of machine translation among language learners: Does it lead to disruptive L2 avoidance? Foreign Language Annals, 57(4), 1094–1114.

55.

Nkhi

S. E.

Shange

(2024). The impact of pedagogical translanguaging in enhancing communicative competence of university students in Lesotho. International Journal of Language Studies, 18(1), 29–52.

56.

Otheguy

García

Reid

(2019). A translanguaging view of the linguistic system of bilinguals. Applied Linguistics Review, 10(4), 625–651. https://doi.org/10.1515/applirev-2018-0020

57.

A. W.

M. M.

Lee

J. C. K.

(2024). Learning and communication in online international higher education in Hong Kong: ICT-mediated translanguaging competence and virtually translocal identity. Journal of Multilingual and Multicultural Development, 45(5), 1732–1745.

58.

Palfreyman

D. M.

Al-Bataineh

(2018). ‘This is my life style, Arabic and English’: Students’ attitudes to (trans) languaging in a bilingual university context. Language Awareness, 27(1–2), 79–95. https://doi.org/10.1080/09658416.2018.1431244

59.

Peker

Arslan

(2020). A critique of Merrill Swain’s output hypothesis in language learning and teaching. Eğitimde Kuram ve Uygulama, 16(1), 99–108. https://dergipark.org.tr/tr/pub/eku/issue/55567/737595

60.

Peker

Özkaynak

(2020). A critical review on the equivocal definition of comprehensible input and the misleading use of the term “acquisition”. Journal of Language Education and Research, 6(1), 238–250. https://www.ceeol.com/search/article-detail?id=880885

61.

Perales-Fernández-de-Gamboa

Orcasitas-Vicandi

(2025). Language teaching in the 21st century: Incorporating culturally sustaining pedagogies for social and cognitive justice in education. Applied Linguistics Review, 16(3), 1403–1427. https://doi.org/10.1515/applirev-2024-0208

62.

Piske

Young-Scholten

(Eds.). (2009). Input matters. Multilingual Matters.

63.

Prilutskaya

(2021). Examining pedagogical translanguaging: A systematic review of the literature. Languages, 6(4), 1–18. https://doi.org/10.3390/languages6040180

64.

Ramirez

P. C.

Salinas

(2021). Reimagining language space with bilingual youth in a social studies classroom. Bilingual Research Journal, 44(4), 409–425. https://doi.org/10.1080/15235882.2021.1994054

65.

Resende

Way

(2021). Can Google Translate rewire your L2 English processing? Digital, 1(1), 66–85.

66.

Sah

P. K.

Kubota

(2022). Towards critical translanguaging: A review of literature on English as a medium of instruction in South Asia’s school education. Asian Englishes, 24(2), 132–146. https://doi.org/10.1080/13488678.2022.2056796

67.

Scarcella

Perkins

(1987). Shifting gears: Krashen’s input hypothesis. Studies in Second Language Acquisition, 9(3), 347–353. https://doi.org/10.1017/S0272263100006720

68.

Schwartz

(2018). The history and promise of machine translation. Innovation and Expansion in Translation Process Research, 43, 161–190.

69.

Sedgwick

(2013). Convenience sampling. BMJ, 347, Article f6304.

70.

Sin-wai

(2014). The development of translation technology 1967–2013. In Sin-wai

(Ed.), Routledge encyclopedia of translation technology (pp. 2–31). Routledge.

71.

Stapleton

C. D.

(1997). Basic concepts in exploratory factor analysis (EFA) as a tool to evaluate score validity: A right-brained approach [Paper presentation]. Annual Meeting of the Southeast Educational Research Association, Austin, TX, United States. https://files.eric.ed.gov/fulltext/ED407419.pdf

72.

Stapleton

Kin

B. L. K.

(2019). Assessing the accuracy and teachers’ impressions of Google Translate: A study of primary L2 writers in Hong Kong. English for Specific Purposes, 56, 18–34.

73.

Ting

F. K.

Tan

K. H.

(2021). Enhancing English Language Vocabulary Learning among Indigenous Learners through Google Translate. Journal of Education and E-Learning Research, 8(2), 143–148.

74.

Tsai

S. C.

(2022). Chinese students’ perceptions of using Google Translate as a translingual CALL tool in EFL writing. Computer Assisted Language Learning, 35(5–6), 1250–1272.

75.

Tsang

(2023). “The best way to learn a language is not to learn it!”: Hedonism and insights into successful EFL learners’ experiences in engagement with spoken (listening) and written (reading) input. TESOL Quarterly, 57(2), 511–536. https://doi.org/10.1002/tesq.3165

76.

Tudor

Hafiz

(1989). Extensive reading as a means of input to L2 learning. Journal of Research in Reading, 12(2), 164–178. https://doi.org/10.1111/j.1467-9817.1989.tb00164.x

77.

Waluyo

Kusumastuti

(2024). Generative AI in student English learning in Thai higher education: More engagement, better outcomes? Social Sciences & Humanities Open, 10, 1–10. https://doi.org/10.1016/j.ssaho.2024.101146

78.

Wang

(2019). Translanguaging in Chinese foreign language classrooms: Students and teachers’ attitudes and practices. International Journal of Bilingual Education and Bilingualism, 22(2), 138–149.

79.

Wang

(2022). Relooking at the roles of translanguaging in English as a Foreign Language classes for multilingual learners: Practices and implications. Frontiers in Psychology, 13, Article 850649.

80.

Wei

(2018). Translanguaging as a practical theory of language. Applied Linguistics, 39(1), 9–30.

81.

Wei

(2023). Towards critical translanguaging pedagogy–new ways of raising and answering questions from a Global South perspective. Critical Inquiry in Language Studies, 20(3), 298–303. https://doi.org/10.1080/15427587.2023.2240457

82.

Williams

(1994). Arfarniad o ddulliau dysgu ac addysgu yng nghyddestun addysg uwchradd ddwyieithog [An evaluation of teaching and learning methods in the context of bilingual secondary education] [Unpublished doctoral dissertation, University of Wales].

83.

Yasar Yuzlu

Dikilitas

. (2022). Translanguaging in the development of EFL learners’ foreign language skills in Turkish context. Innovation in Language Learning and Teaching, 16(2), 176–190.

84.

Yuvayapan

(2019). Translanguaging in EFL classrooms: Teachers’ perceptions and practices. Journal of Language and Linguistic Studies, 15(2), 678–694.

85.

Zaim

Arsyad

Waluyo

Ardi

Al Hafizh

Zakiyah

Syafitri

Nusi

Hardiah

(2024). AI-powered EFL pedagogy: Integrating generative AI into university teaching preparation through UTAUT and activity theory. Computers and Education: Artificial Intelligence, 7, 1–12. https://doi.org/10.1016/j.caeai.2024.100335

Beyond Teacher-Led Approaches: Student-Initiated Translanguaging With Artificial Intelligence Tools in Foreign Language Acquisition

Abstract

Plain language summary

Keywords

Introduction

Literature Review

Conceptual Framework: Translanguaging Theory and Optimal Input Hypothesis

Translanguaging Theory

Optimal Input Hypothesis

AI Translation Tools in Student Foreign Language Learning

Translanguaging and Foreign Language Communicative Competence

The Study

Method

Research Design, Context, and Participants

Teaching Materials and Student Translanguaging Space

Data Collection: Instruments, Measures, Reliability, and Validity

Data Analysis

Results

Descriptive Statistics

Inferential Statistics

Thematic Analysis

Impacts on the Productive Skill

Discussion and Implications

AI Translation Tools in Student Translanguaging Behaviors

Impacts of Using AI Translation Tools Students’ Communication Skills

Conclusion, Limitation, and Recommendation

Footnotes

ORCID iDs

Ethical Considerations

Funding

Declaration of Conflicting Interests

Data Availability Statement

References