Microsoft Word - 6831 5672 Copyright © 2025 The Author IDEAS is licensed under CC-BY-SA 4.0 License Issued by English study program of IAIN Palopo IDEAS Journal on Language Teaching and Learning, Linguistics and Literature ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) Volume 13, Number 2, December 2025 pp. 5672 - 5685 Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective Reihayyu Dwi Cahyani1, Syamdianita2, Aridah3, Weningtyas Parama Iswari4, Ichi Ahada5 1,2,3,4,5 English Education Department, Mulawarman University, Samarinda, Kalimantan Timur Corresponding E-Mail: reihayyudwi@gmail.com Received: 2025-05-28 Accepted: 2025-10-25 DOI: 10.24256/ideas.v13i2.6831 Abstract This study investigates University students' perceptions of the machine translation tool DeepL in the context of learning English as Foreign Language (EFL), involving four participants (two frequent users anda two infrequent users). The main objective was to explore the frequency of use with the tool through the TAM Theory. Using a descriptive qualitative research approach, the purpose of sampling was used to select four university EFL students, who were categorized as frequent and infrequent users of DeepL. Data was collected through semi-structured interviews, and questionnaires. The results showed that frequent users of DeepL appreciated its effectiveness in vocabulary acquisition, translation accuracy, and academic writing support. In contrast, infrequent users expressed concerns about limitations such as the lack of a paraphrasing feature and the formality issues. These findings highlight significant differences in user experience based on frequency of use. This study is among the first to examine differences in user perceptions of DeepL based on frequency of use within the TAM framework, addressing a gap in prior MT research and providing insights into how usage patterns influence perceived usefulness and ease of use. Recommendations for future research include incorporating feedback from users to improve functionality and meet the evolving needs of EFL learners. This research contributes to the understanding of the role of machine translation technology in language learning and offers insights for future research. Keywords: DeepL, English as Foreign Language, Machine Translation, Technology Acceptance Model, User experience, Qualitative Research, University Students. IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5673 Introduction Background The evolution of artificial intelligence (AI) has transformed various domains of human activity, including education (Russell & Norvig, 2020). In the field of English as a Foreign Language (EFL) learning, AI integration has enabled students to overcome challenges in understanding and producing the target language, fostering more efficient learning processes. One of the most influential AI powered tools in this context is Machine Translation (MT). MT tools, such as Google Translate and Microsoft Translator, have become popular due to their ability to enhance vocabulary learning, improve writing, and aid in reading comprehension (Zuhairo & Kembaren, 2024). Machine Translation (MT) using computer software to convert text from a source language to a target language. Machine translation is defined as an automated system that translates between natural languages (Hutchins & Somers, 1992, as cited in Borsatti & Blanco Riess, 2021; Arnold et al., 1994, as cited in Sipayung, 2021), including with or without human involvement (Andriola, 2024). In the context of EFL learning offers several benefits. Students being able to understand class material, improve performance, and develop specialized skills such as vocabulary and grammar (Utimadini, 2023). There are many studies that show students' increased appreciation of this tool. Prayoga (2022) highlighted the value of instant translation and ease of use, while (Safitri, Dewi, & Ramadhan, 2024) found that DeepL was preferred over Google Translate or ChatGPT because of its editing features. (O’Neill, 2012, as cited in Tsai, 2019) says that users produce fewer grammatical errors, and studies by (Wang & Ke, 2022) confirm that DeepL improves writing quality, reduces errors, and supports the revision process in academic contexts. Literature Review Unlike previous MT systems that use word-by-word translation, Neural Machine Translation (NMT) processes entire sentences, resulting in more accurate and natural-sounding translations (Hutchins & Somers, 1992) and one tool that utilizes this technology is DeepL. (Deguchi, Tamura, & Ninomiya, 2019) developed a translation model that considers the relationship between words in the source and target sentences to produce more accurate translations. (Kamaluddin, Rasyid, Abqoriyyah, & Saehu, 2024) confirmed that DeepL shows superior performance compared to other MT tools, especially in terms of fluency and context accuracy. (Bunga & Katemba, 2024) found that DeepL is more effective than Google Translate. In addition, Asmara and Kembaren (2024) highlighted that DeepL is beneficial in completing academic tasks as it can maintain context coherence between the source text and the translated result. Raben (2024) also reported that students found DeepL helpful in translating academic texts from Indonesian to English, which resulted in an overall improvement in translation quality. Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5674 This research uses the Technology Acceptance Model (TAM) as the framework to investigate students’ acceptance of DeepL, a model proposed by Davis (1989) a key determinant that influences user acceptance of a technology. This framework can be used to analyze student usage behavior when using DeepL. Kamaluddin et al. (2024) revealed that attitudes towards this application were influenced by the frequency of using that application and perceived academic utility of that application. The general users are also likely to have more trust in its accuracy and usability while occasional users are more conservative in their opinions. This was also supported by Asmara & Kembaren (2024) who demonstrated that the positive user perceptions were significantly affected by the users' prior experience with DeepL in an academic context. Research Gap Previous studies on DeepL have mostly focused on evaluating the quality of translation output, including lexical and grammatical accuracy, or comparing DeepL with other machine translation tools such as Google Translate. In contrast, this study focuses on students' perceptions of the usefulness and ease of use of the tool in translation tasks. Additionally, while some studies have explored students' general perceptions of DeepL, no study has investigated how these perceptions vary based on frequency of use, an important factor that can influence user experience. This study also employs the Technology Acceptance Model (TAM), a theoretical framework that has been rarely applied in previous research related to DeepL. Therefore, the contribution of this study lies in addressing two specific gaps: (1) the lack of investigation into user perceptions based on frequency of use, and (2) the limited theoretical foundation in previous studies on DeepL perceptions. In line with these objectives, this study generates the following research questions: (1) How do frequent and infrequent student users perceive the usefulness and ease of use of DeepL in English translation? (2) What challenges and future usage intentions do students have when using DeepL, based on their frequency of use? Hence, the main purpose of this study is to explore university students' perceptions of DeepL in terms of its practicality and usefulness in translation tasks, as well as to examine how frequency of use affects their experiences and expectations. The novelty of this study lies in its focus on DeepL with frequency of use and the underlying TAM theory to provide a better understanding of how students interact with machine translation technology. Method This study uses a descriptive qualitative research approach to investigate students' opinions on the use of DeepL Translator in English translation assignments. The focus was on gaining an in-depth understanding of participants’ experiences, attitudes, and perceived challenges. Although a short questionnaire IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5675 was administered, the study remained qualitative in nature because the questionnaire was used solely for descriptive purposes — specifically to categorize participants as frequent or infrequent users and to provide contextual background for the interviews. No statistical hypothesis testing or inferential analysis was conducted, and the primary data source was the semi-structured interviews. This approach aligns with recommendations in qualitative research that allow the use of supplementary instruments, such as surveys, for participant profiling and triangulation without changing the core qualitative orientation (Creswell, 2013). The research participants were four undergraduate students in the English as a Foreign Language (EFL) program, aged between 21 and 22, all of them studying English Education at Mulawarman University. Their English proficiency was based on the completed minimum of four semesters of English courses. Purposive sampling was used to select students who had experience using DeepL for academic translation. The small sample size was chosen to allow in-depth exploration of individual perspectives while keeping the analysis manageable, in line with Creswell (2013) recommendations for qualitative case studies. Two main instruments were used to collect data: semi-structured interviews and a short questionnaire. The questionnaire consisted of five questions. After completing the questionnaire to determine frequency of use, students were interviewed to explore their experiences with DeepL in greater depth. Participants were divided into frequent and infrequent users based on the questionnaire results. Frequent users were defined as those who used DeepL more than three times per week, while infrequent users were those who used it Less than twice a week. The Technology Acceptance Model (TAM), created by Davis in 1989, is the foundation for this study. It places a strong emphasis on Perceived Usefulness (PU) and Perceived Ease of Use (PEOU), or the perceived advantages and simplicity of utilizing the technology. Along with these two elements, this study also examined the difficulties students encountered (Challenges) and their Behavioral Intention (BI) to keep using DeepL. The acquired data was examined using a combination of theme analysis to uncover interview data and descriptive statistics to describe frequency and usage patterns from the questionnaire. This method made it possible to fully comprehend how students felt about using DeepL in an academic EFL setting. Ethical considerations were addressed by obtaining informed consent from all participants before data collection. Participants were assured of confidentiality, anonymity through pseudonyms, and voluntary participation. Data collection took place between March 2025 and May 2025. Results 1. Perceived Usefulness (PU) Here are some of the following excerpts from participants' perceptions of Deepl's advantages, especially when it comes to increasing their vocabulary, improving Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5676 translation accuracy, and completing academic writing tasks are illustrated below. • Vocabulary Learning Support "Kayak nyari tahu vocab aja... kalau nemu vocab yang nggak tahu, aku bakal ke DeepL." (P1, R6) "Nah ternyata pas aku cari di DeepL, artinya 'tempat untuk berbaring'. jadi aku tahu 'bed-ridden' dari situ."(P1, R 20) “It's like finding out vocab... if I find a vocab that I don't know, I will go to DeepL.” (P1, R6) “Well it turns out when I looked it up on DeepL, it means 'a place to lie down.' So, I know 'bed-ridden' from there.” (P1, R20) One of the main benefits of DeepL recognized by the students was its ability to support vocabulary learning. As Participant 1 noted DeepL is the tool used “It's like finding out vocab... if I find a vocab that I don't know, I'll go to DeepL,” showing that students rely on it for immediate vocabulary discovery. In another experience, Participant 1 w shared that, “Well it turns out when I looked up DeepL, it means ‘a place to lie down.’ So, I know 'bed-ridden' from there” (P1, R20), showing how DeepL contributes to vocabulary acquisition incidentally. This theme appeared from a frequent user, suggesting that vocabulary learning is more highly valued by frequent users rather than infrequent. • Translation Accuracy "DeepL lebih akurat penggunaannya... bisa dipilih sesuai konteks." (P2, R23) "Hasil terjemahannya akurat dan mudah dimengerti." (P3, R9) “DeepL is more accurate to use... can be chosen according to context.” (P2, R23) “The translation is accurate and easy to understand.” (P3, R9) In addition to supporting vocabulary learning, students also considered the advantages in terms of translation accuracy, especially compared to other translators. Participant 2 stated, “DeepL is more accurate to use. can be chosen according to context” (P2, R23), while Participant 3 emphasized that “The translation is accurate and easy to understand.” indicating that DeepL provides more than just the meaning of the word as it was originally intended, it offers word options that fit the overall meaning of the sentence. All four participants mentioned translation accuracy (2 frequent & 2 infrequent), but frequent users highlighted contextual fit, while infrequent users highlighted clarity and comprehensibility. This shows how DeepL provides grammatically correct translations that are easily understood by users. • Academic Utility “DeepL untuk keperluan akademik seperti menerjemahkan ke dalam bahasa Inggris atau bahasa Inggris yang sulit saya pahami seperti di artikel atau di jurnal.” (P2, R2) IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5677 "If I want to be more formal, like in academic writing, maybe I use DeepL." (P4, R9) “DeepL for academic purposes, such as translating into English that is difficult for me to understand, such as in articles or journals.” (P2, R2) "If I want to be more formal, like in academic writing, maybe I use DeepL." (P4, R9) DeepL is also useful for academic writing when students need to use formal language and be more structured. Participant 4 said, “If I want to be more formal, like in academic writing, maybe I use DeepL.” Similarly, Participant 2 (CAM) mentioned using DeepL for academic purposes such as translating articles and journals into English, highlighting its role in supporting formal academic communication (P2, R2). This indicates that DeepL can producing structured, formal, and academic style texts. Two participants from frequent and infrequent users referred to this academic benefit, showing it is valued across both user types. 2. Perceived Ease of Use (PEOU) Participants' appreciation for the user-friendly interface, and its uncomplicated components that enhance the speed and convenience of translation is highlighted. The following extract indicates how students evaluated the simple use of DeepL in their daily practice. • Speed and Accessibility "DeepL salah satu yang tercepat untuk dibuka." (P4, R3) "Tinggal tempel teks, langsung diterjemahkan." (P3, R17) “DeepL is one of the fastest to open.” (P4, R3) “Just paste the text, it will be translated immediately.” (P3, R17) Participant 4 stated, “DeepL is one of the fastest to open,” and Participant 3 added emphasizing speed of access as one of the main advantages, “Just paste the text, translate immediately,” indicating that the usage process does not require complicated steps and is very practical for daily learning activities. Two participants from both infrequent users highlighted speed and accessibility, showing that quick access is a stronger motivator for those who use DeepL less often. • Customization Features "Kosa katanya bisa diatur sesuai kalimat yang diinginkan." (P2, R12) "Bisa paraphrase dengan menekan satu kata." (P1, R28) “The vocabulary can be adjusted according to the desired sentence.” (P2, R12) “Can paraphrase by pressing one word.” (P1, R28) Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5678 In addition to easy access and speed, participants appreciate the flexibility that DeepL offers. Participant 2 stated, “The vocabulary can be adjusted according to the desired sentence,” indicating that students believe they are in control of the translation. This feature offers learning vocabulary in a more flexible context. Similarly, participant 1 added, “Can paraphrase by pressing one word.” This shows that the synonym selection function on DeepL helps students with a variety of ideas in editing and rewriting sentences. Two participants from both frequent users highlighted this feature because allows users to edit the translation directly or choose alternative words that fit the context of the sentence. 3. Challenges Despite the many benefits perceived by frequent and infrequent users of DeepL, some participants also mentioned many limitations to its use. Some of the challenges relate to functional limitations, translation styles that are not always contextually appropriate, and the efficiency of its use for certain tasks. The following excerpt proceeds to explain the participants perceived barriers when using DeepL. • Feature Gaps "DeepL enggak punya fitur paraphrase." (P4, R23) "Harus mengubah kata per kata untuk paraphrase." (P3, R20) “DeepL does not have a paraphrase feature.” (P4, R23) “Have to change word by word to paraphrase.” (P3, R20) One of the challenges faced by participants was the no automatic paraphrasing feature. DeepL offers a selection of synonyms for a particular word, but some participants felt this feature is not enough to reorganize an entire sentence. Participant 4 stated, “DeepL has no paraphrase feature,” while participant 3 added, “Have to change word by word to paraphrase,” suggesting that the paraphrasing process is still manual and inefficient to use when rewriting longer texts. Two participants, both of infrequent users expressed this concern, and highlighted that the lack of a paraphrase feature is a main obstacle for users who want more diverse features. • Formality Issues "DeepL masih terlalu formal untuk saya." (P4, R5) “DeepL is still too formal for me.” (P4, R5) In addition to feature limitations, some students complained about the linguistic style of the deep translation being too formal for certain contexts, especially when used in daily conversation and non-academic writing. Participant 4 that infrequent user stated, “DeepL is still too formal for me.” Showing that style mismatches are more problematic for casual use than for academic purpose. Although DeepL was praised for its accuracy and contextual understanding, IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5679 Kamaluddin et al. (2024) recognized that there are still difficulties when working on texts that require in-depth background knowledge. And this example is proof that DeepL should improve its translations across different language styles.  Efficiency Barriers "Tidak ingin membuka tab baru untuk DeepL." (P3, R5) "Jadi, di chat GPT itu lebih instan daripada di DeepL.” (P3, R20) “Do not want to open a new tab for DeepL.” (P3, R5) “So, GPT chat is more instant than DeepL.” (P3, R20) Some students mentioned has efficiency barriers when using Deepl, especially with accessibility and cross-platform integration. Some felt the process was a bit complicated compared to other integrated tools. Participant 3 stated, “Don't want to open a new tab for DeepL,” which shows students chose the tool quite accurately based on speed and convenience. This was reinforced by Participant 3 that stated, “So, GPT chat is more instant than DeepL.” suggesting that DeepL is less competitive compared to other multifunctional tools. One infrequent user described this as a drawback, highlighting convenience as a key factor in choosing a tool. 4. Behavioral Intention (BI) Some students expressed their desire to continue using DeepL because they felt they understood and translated the text, while others considered using other applications that suited their needs. The following quotes present different participants' views on the possibility of continuing to use DeepL in their learning activities. • Continued Use "Aku bakal gunain DeepL terus." (P1, R33) "Akan tetap pakai DeepL sampai ada yang lebih akurat." (P2, R21) “I will continue to use DeepL.” (P1, R33) “I will keep using DeepL until there is something more accurate.” (P2, R21) Some students have shown strong intentions to continue using it in depth. For example, participant 1 stated, “I will continue to use DeepL,” while participant 2 said, “I will keep using DeepL until there is a more accurate one,” which suggests that if DeepL still excels in accuracy, students will continue to rely on it. Two frequent users expressed higher intention to use due to accuracy. Therefore, satisfaction with in-depth functionality plays an important role in shaping future usage behaviour. Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5680 • Conditional Use "Kalau ada fitur tambahan, kenapa tidak?" ( P4, R33) “If there are additional features, why not?” (P4, R33) In addition to students who mentioned their intention to continue using the platform, there were participants who communicated their intention to use conditional use. Participant 4 stated, “If there are additional features, why not?” This indicates a desire to continue using DeepL, especially if improvements are made. This participant fell into this category. And the finding is in line with the Technology Acceptance Model (TAM) framework (Davis, 1989), which states that satisfaction and expectations are fundamental and perceived usefulness and ease of use influence intention to use. • Preference for Alternatives "DeepL bukan yang utama... ChatGPT lebih efektif." (P3, R23) “DeepL is not the main one... ChatGPT is more effective.” (P3, R23) Deepl was considered accurate and beneficial by many students, but some participants indicated other translation tools that were considered more flexible and efficient. Participant 3 (Row 23) stated, “DeepL is not the main one... ChatGPT is more effective,” indicating that ChatGPT's multifunctional features and ability to handle different types of tasks made it more attractive to users seeking efficiency in the learning process. This suggests that the intention to continue using DeepL may influence the development of other technologies that provide more comprehensive services on the platform. This highlights that even satisfied users may shift if alternatives offer better integration. Discussion The findings of this study show various aspects of students' experiences with DeepL in academic translation. Each section has been identified and is outlined below in relation to the existing literature and theoretical perspectives. Using the Technology Acceptance Model (TAM) theoretical framework developed by Davis (1989), this study adds to the understanding of how the relationship between of how perceived usefulness (PU), perceived ease of use (PEOU), and behavioral intention (BI) in the context of learning English as a foreign language (EFL) in Indonesia, especially among university students. The students recognized various benefits of DeepL that contributed to their academic tasks. One of the main benefits was its role in supporting vocabulary learning. Frequent users, such as Participant 1, stated, “It's like finding out vocab... if I find a vocab that I don't know, I will go to DeepL.” This experience shows how IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5681 students use DeepL to build their understanding with new vocabulary. The results show that for frequent users, DeepL has become part of their daily or weekly academic routine. This strengthens the perception of usefulness (PU) because it consistently improves vocabulary mastery and the use of contextually appropriate expressions, which in turn encourages their intention to continue using DeepL. This finding supports Laksana and Komara's (2024) conclusion that DeepL helps EFL students acquire vocabulary by providing a variety of options. DeepL is also recognized for its translation accuracy. An infrequent user expressed by an infrequent user, admitting, “The translation is accurate and easy to understand” These statements reflect confidence in DeepL's ability to produce accurate and easy to understand translations. Kamaluddin et al. (2024) also found that DeepL produced more natural translations and fewer errors compared to other tools. In an academic setting, even users who use DeepL less frequently see the value of DeepL in producing formal and structured writing. This is in line with the findings of Asmara and Kembaren (2024) who found that students rely on DeepL when working on academic assignments such as theses or journals. However, for infrequent users, high PEOU (Perceived Ease of Use) and accuracy alone are not enough to keep BI, indicating that feature completeness may be more influential than ease of use. Most students found DeepL to be fast, accessible, and user-friendly. Infrequent user noted that it is “one of the fastest to open” and frequent uses valued DeepL’s customizable translation feature. This statement shows that the speed and ease of use of DeepL is the main thing. Laksana and Komara (2024) also found that EFL students appreciated the ease of DeepL because it helped them save time in doing academic work. Students liked DeepL's translation customization feature, which allowed them to adjust the choice of words in sentences. For frequent users, these features strengthen the relationship between PEOU and PU, which directly influences continued use. There are also some limitations to this platform in terms of translation. Infrequent users indicate that although DeepL provides various suggestions on words, it does not have paraphrases on whole sentences, making it less efficient for revision tasks. Kirana et al. (2024) emphasize that such limitations reflect the broader need for innovation in educational technology. And one of infrequent user highlighted DeepL's tendency to be too formal, which aligns with Kamaluddin et al. (2024) who note that DeepL still struggles in tasks that require cultural nuances or stylistic variations. In the context of teaching English as a foreign language (EFL) in Indonesia, the use of formal language is often considered an added value and supports academic needs. However, this tendency can also be a limitation when faced with tasks that require creativity or a more casual style of language. Infrequent users also find DeepL less efficient than other platforms that are much more efficient. These comments suggest that infrequent users could use a multifunctional tool that combines features into one accessible interface, which Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5682 DeepL currently lacks. Participants also showed varying intentions to continue using DeepL. Frequent users showed strong satisfaction and commitment. Frequent users showed a high level of trust in DeepL's capabilities, especially in producing accurate translations for academic needs. This aligns with TAM’s proposition that high PU and PEOU correlate to stronger BI. But our findings show that for some infrequent users or lower BI shows that adoption can be external factors such as tool versatility can disrupt. In contrast, infrequent users expressed conditional or hesitant intentions. Participant 4 said, “If there are additional features, why not?” This conditional openness reflects the importance of continuous development in technology to meet evolving expectations. As outlined in the Technology Acceptance Model (TAM) (Davis, 1989), satisfaction, perceived usefulness, and ease of use shape continued usage behavior. So, some users who rarely use them prefer alternative tools. Participant 3 stated, “DeepL is not the main one... ChatGPT is more effective.” This preference suggests that infrequent users may switch to a more multifunctional tool if DeepL does not evolve to meet broader academic and usability needs. Kirana et al. (2024) also observed that DeepL's limitations in handling specialized vocabulary may reduce its long-term appeal in academic contexts. The results of this study show significant differences in how frequent and infrequent users of DeepL perceive the usefulness and ease of use of this application in English translation. Frequent users (participants 1 and 2) consistently reported that DeepL was very useful in learning vocabulary, producing accurate translations, and assisting informal academic writing. They also rated DeepL as easy to use due to its simple interface, quick access and flexible features such as synonym, selection, and context adjustment. In contrast infrequent users (participants 3 and 4) still recognize these benefits, but use DeepL more selectively as they found features less useful, such as having to open new tabs or preferring other tools such as chat GPT that are more multifunctional. A structured comparison between these groups shows that while PU is a key factor for frequent users, adoption by infrequent users is more sensitive to efficiency, integration, and flexibility specific to particular tasks Conclusion This study sought to find out how University students perceive DeepL in learning English as a foreign language (EFL). And the results show that both frequent and infrequent users of this translation tool have some usability and usability issues with DeepL, although with different levels of enthusiasm. By applying TAM as a theoretical framework, this study expands the application of the model in the context of teaching English as a foreign language (EFL). The study shows that although PU and PEOU can generally predict BI, in some cases, feature completeness and relevance to context can change or replace these connections. Limitations of this study include the small and homogeneous sample size (n = 4) which may not fully represent the diverse experience of all EFL learners, and the IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5683 focus only from one university in Indonesia. However, this study contributes to knowledge by highlighting the different ways in which frequency of use affects students' experiences with machine translation technology. In addition, the use of the Technology Acceptance Model (TAM) Theory by Davis 1989 as the theoretical basis in this study also breaks new ground for future research. Based on these results, it is recommended that further research be conducted with a larger and more diverse sample of EFL learners to validate the results. Future studies should also consider using a mixed-methods approach to measure the strength of the relationship between TAM constructs and explore the moderating role of feature richness and cultural expectations. Future research should also investigate the impact of certain features, such as paraphrasing ability, on user satisfaction and learning outcomes. In addition, developers of machine translation tools such as DeepL should consider including feedback from users to improve usability. For educators, the structured use of machine translation (MT) tools in classroom activities, such as translation comparison analysis, can help improve critical language awareness. This would not only improve the user experience, but also ensure that the tool helps the growing needs of EFL learners in various academic contexts. Acknowledgement The author expresses gratitude and thanks to Allah SWT for the strength and blessings that have helped and facilitated the completion of this research properly and as it should be, as well as enabling the author to complete this research project. Thanks, are also extended to the supervisors for their guidance, useful advice, and valuable direction at every stage of this research. The author also extends gratitude to family, father, mother, and two brothers who have consistently supported and encouraged the author throughout the process of conducting this research. To the author’s friends, college and school friends, and special friend, who have always been a source of support. The encouragement and advice from these beloved individuals have played a significant role in assisting the research process. References Alhaisoni, E., & Alhaysony, M. (2017). An Investigation of Saudi EFL University Students’ Attitudes towards the Use of Google Translate. International Journal of English Language Education, 5(1), 72. https://doi.org/10.5296/ijele.v5i1.10696 Asmara, D. S. M., & Kembaren, F. R. B. (2024). Student’s perception towards the use of DEEPL Translator in writing thesis or journal for English education students. IJLECR - INTERNATIONAL JOURNAL OF LANGUAGE EDUCATION AND CULTURE REVIEW, 10(1), 117–126. https://doi.org/10.21009/ijlecr.v10i1.47937 Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. Reihayyu Dwi Cahyani, Syamdianita, Aridah, Weningtyas Parama Iswari, Ichi Ahada Understanding EFL Students’ Acceptance of DeepL for English Translation: A Technology Acceptance Model Perspective 5684 https://doi.org/10.1191/1478088706qp063oa Borsatti, D., & Riess, A. B. (2021). Using machine translator as a pedagogical resource in English for specific purposes courses in the academic context / O uso do tradutor automático como recurso pedagógico na aula de inglês para propósitos especı́ficos no contexto acadêmico. Revista De Estudos Da Linguagem, 29(2), 829–858. https://doi.org/10.17851/2237-2083.29.2.829- 858 Bunga, E. L. M., & Katemba, C. V. (2024). COMPARING TRANSLATION QUALITY: GOOGLE TRANSLATE VS DEEPL FOR FOREIGN LANGUAGE TO ENGLISH. EDUSAINTEK JURNAL PENDIDIKAN SAINS DAN TEKNOLOGI, 11(3), 1147– 1171. https://doi.org/10.47668/edusaintek.v11i3.1264 Creswell, J. W. (2013). Qualitative Inquiry and Research Design: Choosing among five approaches (3rd ed.). Sage Publications Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319. https://doi.org/10.2307/249008 Deguchi, H., Tamura, A., & Ninomiya, T. (2019). Dependency-Based Self-Attention for Transformer NMT. In Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2019) (pp. 239–246). INCOMA Ltd. https://doi.org/10.26615/978-954-452-056-4_028 Hutchins, W. J. (1986). Machine Translation: Past, Present, Future. Ellis Horwood. Hutchins, W. J., & Somers, H. L. (1992). An Introduction to Machine Translation. Academic Press. Ismayanti, D., Said, Y. R., Usman, N., & Nur, M. I. (2024). The Students Ability in Translating Newspaper Headlines into English A Case Study. IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature, 12(1), 108-131. Kamaluddin, M. I., Rasyid, M. W. K., Abqoriyyah, F. H., & Saehu, A. (2024). Accuracy analysis of DeepL: Breakthroughs in machine Translation technology. Journal of English Education Forum (JEEF), 4(2), 122–126. https://doi.org/10.29303/jeef.v4i2.681 Laksana, N. K. N., & Komara, N. C. (2024). Indonesian EFL Students’ Perceptions of DEEPL Machine Translation Tool: Utilization, Advantages, and Disadvantages. Journal of Language and Literature Studies, 4(2), 256–276. https://doi.org/10.36312/jolls.v4i2.1931 Masruddin, M., & Nasriandi, N. (2022). Lexical and Syntactical Errors Performed by Junior High School Student in Writing Descriptive Text. IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature, 10(1), 1094-1100. Munawwarah, M., & Martriwati, ; (2024). STUDENTS’ PERCEPTION OF USING DEEPL AS MACHINE TRANSLATION IN ENGLISH LEARNING. 5(2), 284–295. https://doi.org/10.22236/ellter.v5i2.15813.g4779 Prayoga, G. B. (2022). Students’ perception of using machine translation tools in the IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5685 EFL classroom. UICELL Conference Proceedings, (6). https://journal.uhamka.ac.id/uicell/article/view/11049 Raben, S. K., Sirande, N., Arrang, J. R. T., & Ilham, M. (2024). A comparative analysis of the accuracy of machine translation in translating English to Indonesian. Teaching English as a Foreign Language Overseas Journal, 3(1), 1–10. https://doi.org/10.47178/t0gw3271 Rustika, P. (2024). Students’ perceptions of the Use of deepL translator in translating academic text from Indonesian into English: A Case Study at State Islamic University of Sunan Gunung Djati Bandung - Digital Library UIN Sunan Gunung Djati Bandung. (n.d.). Retrieved from https://digilib.uinsgd.ac.id/102629/ Safitri, B., Dewi, U., & Ramadhan, A. (2024). EFL Students’ Preferences of Artificial Intelligence (AI) for Writing. Jurnal Ilmu Sosial, Humaniora Dan Seni, 3(1), 702–705. https://doi.org/10.62379/jishs.v3i1.2008 Tsai, S. (2019). Using google translate in EFL drafts: a preliminary investigation. Computer Assisted Language Learning, 32(5–6), 510–526. https://doi.org/10.1080/09588221.2018.1527361 Utimadini, N. J. (2023, December 30). Exploring perceptions of machine translation as a tool for EFL learning. Retrieved from https://journal.ikippgriptk.ac.id/index.php/bahasa/article/view/7128 Wang, J., & Ke, X. (2022). Integrating Machine Translation into EFL writing instruction: process, product and perception. Journal of Language Teaching and Research, 13(1), 125–137. https://doi.org/10.17507/jltr.1301.15 Zuhairo, Z., & Kembaren, F. R. W. (2024). Intermediate Students’ perceptions of the transformation of online translation Engine. IJLECR - INTERNATIONAL JOURNAL OF LANGUAGE EDUCATION AND CULTURE REVIEW, 10(1), 12–20. https://doi.org/10.21009/ijlecr.v10i1.45004