© hamonazaryan1 / Pixabay
Език и култура
INTEGRATING CHATGPT INTO ENGLISH-UKRAINIAN TRANSLATOR TRAINING
https://doi.org/10.53656/for2026-03-08
Резюме. The article examines the ways of integration of ChatGPT into translator training at the bachelor’s level and presents the results of the experiment conducted during 2023-2025 among the 3rd-year students of Ternopil Volodymyr Hnatiuk National Pedagogical University. Methods included a pedagogical experiment to examine the integration of ChatGPT into translator training; semi-structured interviews to explore students’ experiences; analysis of students’ translation assignments completed with and without ChatGPT; classroom observation; qualitative and quantitative analysis of the collected data. Findings indicate that most of the students see the potential of ChatGPT in their professional training. The obtained data was used to create the activities for the students. The study highlights the need to integrate AI-assisted tools into translator training to meet the expectations of the modern labour market, carefully choosing the assignments, giving clear and strict instructions for ChatGPT, adjusting the principles of academic integrity and taking into consideration the needs of students.
Ключови думи: ChatGPT; translator training; AI-assisted tools; labour market; academic integrity
Introduction
Entering the era of AI, humanity faces new challenges in the sphere of labour market. In this context, ChatGPT is attracting increasing interest of language professionals and translators, who explore the potential of this tool, its strengths and limitations, in the attempt to estimate and predict the influence of the technology on their jobs.
ChatGPT has received much attention of scholars since its release for public use in 2022. Within the next few years, ChatGPT is likely to become an important tool in translation industry. Hence, the approaches to professional training of translators should be revised and adapted. Few researchers have addressed this issue so far.
There is still a need for discussion on English-Ukrainian machine translation performance by ChatGPT. The fact that English is a West Germanic language, whereas Ukrainian is an East Slavic language (although both belong to the Indo-European family of languages), can pose significant challenges to the chatbot. Jiao et al (2023) conclude that ChatGPT performs competitively with commercial translation products on high-resource European languages. However, the researchers claim that it lags behind on low-resource or distant languages. In our study, we focus on English-Ukrainian translation, which is of particular interest due to the fact that Ukrainian falls into the category of medium-resource languages (on the basis of data ratios in the CommonCrawl corpus).
Hence, the aim of our work is to broaden current knowledge of ChatGPT failures in English-Ukrainian translation and develop the ways of incorporating this tool in the process of professional training of English-Ukrainian translators at the bachelor’s level, making the most of this large language model.
According to the objective of this paper, the solution for the following fundamental tasks is offered:
1) to reveal the need to incorporate ChatGPT into professional training of future translators;
2) to design and conduct a pedagogical experiment in Ternopil Volodymyr Hnatiuk National Pedagogical University;
3) to analyse the main mistakes made by ChatGPT in the process of English-Ukrainian translation;
4) to provide the examples of activities for future translators at the bachelor’s level including the use of ChatGPT and post-editing.
Literature Review
AI has many applications in the field of education, in particular foreign language teaching. Recent developments in AI technologies have led to their increased use in translation industry. Though, the previous work of scholars has been limited in the number of languages under analysis. Very few Ukrainian scholars address the issues of embedding ChatGPT into curriculum. The areas under consideration are the following: incorporating AI for teaching translation as a task generator (Khatser, 2024) and as a pre-production, production and post-production assistant (Karaban & Karaban, 2025, pp. 95 – 99). However, no one to the best of our knowledge has studied the use of ChatGPT in teaching English-Ukrainian translation with an in-depth error analysis of subject-field texts and self-designed post-editing activities.
Authors engage in analyzing the compatibility of the use of Chat GPT with the norms of academic integrity (Ostapovych et al, 2023; Kohnke et al, 2023). They underline the necessity to draw up guidelines for using this tool in educational settings. The author of A Guide to Conversational AI suggests using ChatGPT for grammar checking, proofreading, editing, and text summarization (Atlas, 2023). To reveal the potential of this large language model, Atlas (2023) recommends adhering to the following principles of prompting: choose your words carefully, define the conversation with purpose and focus, be specific and concise, ask for more information, and keep the conversation on-track.
A growing body of literature has studied the use of ChatGPT for language teaching (Kohnke et al, 2023). In their analysis, Kohnke et al reveal the need for the development of the students’ and teachers’ digital competencies, i.e. provide ideas for how to utilize ChatGPT pedagogically and demonstrate the importance of fact-checking the content produced by the chatbot. The researchers conclude that the following forms of digital competence are needed to use ChatGPT for educational purposes efficiently: technological proficiency, pedagogical compatibility, and social awareness.
Experiments on the use of ChatGPT were conducted in 2023 by the group of researchers (Peng et al) who aimed to make the most of this tool for machine translation by revisiting several aspects such as temperature, task information, and domain information. They revealed a correlation between lower temperature and better performance. In addition, emphasizing the task information and introducing domain information can further improve the tool’s performance. Meanwhile, chain-of-thought prompting leads to word-by-word translation behavior, which degrades the chatbot’s performance.
In the article on ChatGPT for Arabic-English translation, Khoshafah (2023) mentions the limitations in the use of ChatGPT for translation of medical reports, scientific studies, literary works, and legal documents. Besides cultural differences, the researcher mentions the following challenges for this translation-related tool: the issue of word order, difficulties in translating pronouns, idiomatic expressions, colloquial terms, and homonyms.
Larroyed in 2023 was among the first to explore the use of ChatGPT in patent translation. She states that ChatGPT and Patent Translate can be successfully employed by patent professionals as complementary tools. Meanwhile, to ensure the accuracy of machine translation, it is necessary to devise a strategy that involves a quality human review.
Orasan (2023) discusses scenarios in which ChatGPT can prove useful for translators: a monolingual writing support tool, a translation engine, an evaluation metric (to rate the quality of translation), and a tool for terminology extraction. A recent review of the literature on exploring the potential of large language models in translation (Siu, 2023) found that ChatGPT can be employed to support the following translation tasks: contextual clarification of expressions; cultural explanation of expressions; explanation of technical terminology and simplification of complex text; draft translations and inspiration; error detection, grammar checking and quality assessment; editing and stylistic recommendations; interactive translation and editing with progressively improving results. Among the strengths of this AI technology, Siu (2023) mentioned speed and cost, versatility and flexibility, creativity and adaptability in translation, translation with contextual awareness. According to the researcher’s findings, the chatbot’s limitations in the context of English-Chinese translation are quality, hallucination and bias, limited support for Chinese language, output variability, security and stability concerns. Meanwhile, Ostapovych et al draw our attention to common ChatGPT English-Ukrainian translation errors: difficulties in term translation, violations of the stylistic norms of the target language, the use of russianisms. Karaban et al (2024) investigate translations of Ivan Franko’s poems done by the translator Percival Cundy and the GPT-3.5 AI language model. On the basis of quantitative and qualitative metrics analysis, the scholars come to the conclusion that the above-mentioned chatbot’s translations of Ukrainian poetry “demonstrated a caliber cоmparable to human efforts”. Among inaccuracies in AI-driven translations, they point out grammatical complexities, cultural contexts, dialectal forms, and botanical terms.
Research methods
According to the tasks assigned and in order to get detailed and qualitative data about the studied phenomenon, 141 3rd-year students of translation department in Ternopil Volodymyr Hnatiuk National Pedagogical University during 2023 – 2025 studying Practice of Translation course took part in the experiment. The reason behind choosing these participants was based on the fact that they have already obtained knowledge in theory of translation, had a course on the contrastive typology of English and Ukrainian, elective translation courses and started their 2-year course in practice of translation. Data collection took place through the questionnaire prepared by the authors, designed tasks with the use of ChatGPT-3.5 and further analysis as well as pedagogical observation and open class discussions. For our research on translation practices, ChatGPT-3.5 was selected from among the available AI platforms because it is the AI tool most commonly used by our students, due to its accessibility and support for the Ukrainian language. All translations were generated using the free web version of ChatGPT-3.5 via its standard interface. Consequently, it was not possible to specify or adjust the temperature parameter, and all participants completed the translation tasks under the same default system settings. The information obtained was qualitatively and quantitively processed and graphically represented with description. A comprehensive literature review was conducted prior to collecting and analyzing the data.
The results and discussion
Students’ Attitude and Use of ChatGPT
At this stage of the research (2023 – 2024) a questionnaire (with multiple-choice and open-ended questions) was designed and distributed among the 3rd-year students. The participation was anonymous and voluntary.
According to our survey, 91.5% (N = 129) of respondents reported having used ChatGPT, whereas 8.5% (N=12) claimed never using it (Fig. 1). As for the tasks performed with the help of ChatGPT, the students mentioned searching for additional information, filling in the gaps activities, translation, editing, making agenda, tutoring, and doing homework.
Figure 1. Usage of ChatGPT among respondents in 2023 – 2024
As for the need to incorporate ChatGPT into future translators’ training, 90.1% (N=127) were convinced that such integration is necessary, mentioning several reasons, such as: analysing AI-generated translations followed by human post-editing, editing translated texts, finding translation equivalents, searching additional information required for translation, creating translation-based tasks for students, and improving comprehension of the source text. Several of these applications have also been mentioned by Orasan (2023). However, 9.9% (N=14) of the respondents didn’t see any practical value in incorporating this AI tool for their professional training (Fig. 2). This perception may reflect a certain degree of reluctance to adopt emerging tools, sticking to more traditional ones, or concerns about reliability of AI-generated translations.
Figure 2. Perceived need to incorporate ChatGPT into translators’ training
The students were also asked whether their lecturers encouraged them to use ChatGPT to complete specific tasks as part of their translation training courses. 10.6% (N=15) responded “yes” and 89.4% (N=126) answered “no”. This finding might be explained by the lecturers’ concerns that students will overuse AI and violate academic integrity (Fig. 3). To address this concern, we agree with Ostapovych et al, (2023) and Kohnke et al (2023) that clear institutional guidelines and ethical use of AI are essential. It is worth noting that in 2025 the Policy on the Use of Artificial Intelligence in Scientific, Technical, and Innovative Activities at Ternopil Volodymyr Hnatiuk National Pedagogical University4 was developed and published on the official university website, providing guidelines on the use of AI by lecturers and students.
Figure 3. Students’ reports on lecturers’ encouragement to use ChatGPT
It’s well known that for adequate rendition of a source text, translators often need to search for additional information. Therefore, the participants in the survey were asked about their sources of information during translation tasks. As shown in the results, 53.2% (N=75) reported using the Internet (Google Search), 25.5% (N=36) – dictionaries, 25.5% (N=36) – chat GPT (36), 2.1% (N=3) – books, 2.1% (N=3) – DeepL, and 2.1% (N=3) – Reverso Context. Some of the students mentioned using more than one source of information. Hence, the data reveal that most students rely on the Internet (Google search), ChatGPT, and dictionaries (both physical and electronic) (Fig. 4), which appears logical in the context of advanced digitalisation.
Figure 4. Students’ preferred sources of information for translation tasks
The participants were also asked to distinguish between AI and human translation. They were offered 3 pairs of passages taken from social-political domain and translated from English into Ukrainian. It’s worth mentioning that 91.5% (N=129) could successfully distinguish human rendition from machine translation (Fig. 5). Their choice was explained by the lack of naturalness, weird word choice, some sense mistakes, too many calques, unusual syntax, punctuation mistakes, and stylistic discrepancies.
Figure 5. The rate of distinction between AI and human translation
ChatGPT Failures in English-Ukrainian Translation
In the course of our research (2024 – 2025), we revealed the most common mistakes made by ChatGPT in English-Ukrainian translation. In this part of the article, we will illustrate ChatGPT failures in English-Ukrainian translation of Calculate payroll deductions and contributions. About the deduction of Employment Insurance (EI) premiums by the Government of Canada (2022) 1 used for classroom work during the course of “Practice of Translation”.
For instance, word for word translation results in
– repetition which leads to syntactically ill-formed sentences and should be avoided in Ukrainian due to aesthetic reasons:
If you’re a fisher, barber or hairdresser, or if you drive a taxi or other passenger vehicle, you don’t need to register for the self-employed program.
Якщо ви рибалка, перукар або фризер, або якщо ви водій таксі або іншого пасажирського транспортного засобу, вам не потрібно реєструватися в програмі для самозайнятих осіб.
Moreover, the use of the dialectal word «фризер» in translation is not justified. Meanwhile, it would be more accurate to use the borrowed word «барбер» (transliteration from the English word “barber”) instead of «перукар» to define a person who cuts men’s hair.
– mistakes with prepositions:
<…> people who are away from work because they’re pregnant or have recently given birth.
<…> люди, які відсутні з роботи через вагітність або недавно народили.
– mistakes with pronouns. In Ukrainian, the possessive and reflexive pronouns are omitted if they are considered semantically redundant:
You can apply for special benefits 12 months after your confirmed registration date.
Ви можете подавати заявки на спеціальні вигоди через 12 місяців після вашої підтвердженої дати реєстрації.
In addition, the use of the same root words in the translated sentences results in inadequate rendering of the original style:
If you run your own business or control more than 40% of your corporation’s voting shares, this program can provide you with access to special benefits as early as 12 months after registering.
Якщо ви власник власного бізнесу або контролюєте понад 40% голосів акцій вашої корпорації, ця програма може надати вам доступ до спеціальних вигод уже через 12 місяців після реєстрації.
If you’re a shareholder of a corporation where you work as an employee and you control more than 40% of that corporation’s voting shares, your premiums are based on your employment income.
Якщо ви є акціонером корпорації, де ви працюєте як працівник і контролюєте більш як 40% голосуючих акцій цієї корпорації, ваші внески обчислюються на підставі вашого доходу від праці.
According to their endings, all Ukrainian nouns belong to one of the grammatical genders: masculine, feminine or neuter. Moreover, there are seven cases in Ukrainian that influence the endings of nouns. It is obvious that ChatGPT has difficulties when dealing with these grammatical categories:
<…> parents who are away from work to care for their newborn or newly adopted child.
<…> батьки, які відсутні з роботи, щоб доглядати за своїм новонародженим або новоусиновленим дитином.
Compassionate care.
Скорботний відпустка.
We also observe ChatGPT’s problems when dealing with the category of number:
The Canada Revenue Agency (CRA) will confirm the amount of your premiums based on the self-employed income you report on your tax return.
Канадське агентство доходів (CRA) підтвердить суму вашої внесків на підставі доходу від самозайнятості, який ви вказали у своїй податковій декларації.
In this case, your insurable earnings* from both employment and self-employment could be combined to increase your benefit rate.
У цьому випадку ваші страхові внески з обох праці можуть бути об’єднані для підвищення вашої ставки вигод.
In certain cases, the translation provided by ChatGPT is semantically redundant:
<…> people who provide care or support to a crtically ill or injured person 18 or over.
<…> люди, які надають догляд або підтримку критично хворій або пораненій особі віком від 18 років і старше.
One of the challenges for ChatGPT is translation of abbreviations:
The CRA determines what types of earnings are insurable.
Країнний агентство доходів визначає, які види доходів є страховими.
Агентство з розміщення оголошень визначає, які види доходів є страховими.
In this context, the abbreviation CRA stands for “Canada Revenue Agency” – «Канадське податкове агентство». In both above mentioned examples, we observe incorrect decoding of this lexical item by ChatGPT.
This AI chatbot can make inexplicable linguistic choices and generate incorrect abbreviation in the target text:
Employment Insurance (EI) has a program designed for self-employed people.
Страхування в разі безробіття (ЄС) має програму, призначену для самозайнятих осіб.
Occasionally, an incorrect abbreviation in the target text is the result of transcoding (transliteration or transcription) of the English shortened form of a phrase:
The CRA determines what types of earnings are insurable.
КРА визначає, які види заробітку є страховими.
Our observations indicate that this approach is used when ChatGPT “doesn’t know” the meaning of the lexical unit:
Compassionate care
Компасна допомога
However, in the majority of cases, ChatGPT leaves abbreviations untranslated, as in the following case:
People in these professions should apply for EI benefits as an employee.
Людям цих професій слід подавати заявки на отримання вигод EI як працівники.
Hence, ChatGPT makes systematic errors in translation of abbreviations and is still struggling to effectively maintain context.
In addition, ChatGPT needs more training in the sphere of syntagmatic relations. Occasionally, at this stage of its development, the tool produces English-Ukrainian translation that sounds unnatural:
If you’re a resident of Quebec, sickness, compassionate care and family caregiver benefits are available to you through our program.
Якщо ви є мешканцем Квебеку, вам доступні вигоди у зв’язку з хворобою, співчутливий догляд та доглядальницька допомога для сім’ї через нашу програму.
If you run your own business or control more than 40% of your corporation’s voting shares, this program can provide you with access to special benefits as early as 12 months after registering.
Якщо ви власник свого бізнесу або контролюєте більше 40% голосових акцій вашої корпорації, ця програма може надати вам доступ до спеціальних пільг уже через 12 місяців після реєстрації.
It was found that ChatGPT offers translation of expressions calqued from Russian:
<…> 40 or 69 depending on the option you choose.
<…> 40 або 69 в залежності від обраної опції.
If you have, you’ll continue paying premiums for as long as you’re self-employed.
Якщо ви отримували, ви будете продовжувати сплачувати внески на протязі усього періоду вашої самозайнятості.
EI maternity and parental benefits.
Вигоди вагітним і батькам по програмі страхування зайнятості.
These observations relate to prepositional phrases as well as separate words:
To be eligible for benefits between January 1, 2024 and December 31, 2024, you need to have made at least $8,492 in net self-employed earnings in 2023.
Щоб бути підходящим для отримання виплат між 1 січня 2024 року і 31 грудня 2024 року, вам потрібно було зробити принаймні $8,492 чистих доходів від самозайнятості в 2023 році.
At times, translation provided by AI is completely senseless:
If you’re a fisher, barber or hairdresser, or if you drive a taxi or other passenger vehicle, you don’t need to register for the self-employed program.
Якщо ви рибалка, перукар або перукарня, або ви керуєте таксі або іншим пасажирським транспортом, вам не потрібно реєструватися в програмі для само зайнятих осіб.
According to the Collins Dictionary2, there are two definitions of “hairdresser”: 1. a person who cuts, colours, and arranges people’s hair; 2. a shop where a hairdresser works. Understanding the context helps a human translator determine the meaning of the word and provide accurate translation on the basis of the first definition. Meanwhile, ChatGPT generates its translation based on the second definition.
For the same reason, ChatGPT finds it hard to translate terms. For example, the expression “regular benefits” is translated as «звичайні вигоди», «звичайні виплати», and «звичайні пільги». The Collins Dictionary provides more than 20 definitions for the word “regular” and the first in the list is “normal, customary, or usual”. Obviously, this served as a basis for the translation provided by ChatGPT. However, in this context, the word “regular” means “occurring at fixed or prearranged intervals” and should be translated as «регулярні».
Since any language is a living organism, traditional spelling can be revised. For example, the word «інтернет» must begin with a small letter. Obviously, ChatGPT is unaware of new capitalization rules in Ukrainian:
<…> register online for the program.
<…> зареєструйтеся в Інтернеті для участі в програмі.
Summing up, we clearly see that the most common failures fall into the following categories: (1) lexical and terminological issues (inaccurate terminology, incorrect abbreviation decoding, problems with dialectal words conveying); (2) grammatical (mistakes in prepositions, pronouns, noun endings and capitalisation, number agreement); (3) syntactic (unnatural Ukrainian structure, disrupted syntagmatic relations); (4) stylistic and pragmatic (unnecessary repetition, semantic redundancy, inadequate style rendering, interference from Russian calques and completely nonsensical AI-generated translations). This analysis was later used for honing effective post-editing skills both in the classroom and as part of self-study assignment.
Activities offered for effective post-editing
As we could see in the previous subheading, ChatGPT for now cannot cope with translation tasks successfully, however, it can do part of the work, allowing translators to process more material during the allotted time. Thus, in the modern era of rapid technological advancement, we share the view of Carmo & Moorkens (2020) that post-editing is necessary and “rewarding” for translators “in a demanding professional environment”. It is in line with European Master’s in Translation Competence Framework (2022) 3 which describes 5 main areas of competence including translation and states that students should “know how to …post-edit MT output using style guides and terminology glossaries to maintain quality standards in MT-enhanced translation projects” (ibid, p.8). In order to develop the needed skill, specially designed activities were offered to the students. Let’s consider the most productive ones with the suggestions for their practical use for students of different proficiency levels and group sizes.
Activity 1. Read the text. Make a list of unknown vocabulary items, abbreviations and new or confusing for a translator concepts. Using ChatGPT, make a table with explanation, suggested translation and identified sources of information. Verify the AI-generated information, paying special attention to accuracy and reliability. Prepare the edited table and share with your groupmate. In pairs, make one table and present it in open class.
This exercise can be offered as a part of the preparatory stage for the translation. However, the lecturer should emphasize the importance of providing simple, strict, clear and detailed instructions for ChatGPT (Khatser, 2024) in order to obtain better results. Students may work either in pairs or in small mixed-ability groups depending on the level and learning objectives of the class.
Activity 2. Translate the text using ChatGPT and then translate it on your own (Fig. 6). Compare your translation with the one done by ChatGPT and single out lexical, grammatical mistakes and mistakes that significantly change the meaning of the source text, make a conclusion (Fig. 7).
Figure 6. A snapshot of a submitted task with the translation done by ChatGPT and by one of the students
Figure 7. A snapshot of a submitted task with student’s analysis and conclusions
The offered activity can be done individually, in pairs or in small groups. The students may work on this task in the classroom or at home. Additionally, a lecturer may provide students with lists of possible mistakes and, if needed, examples. The activity can be modified and targeted only on one specific type of issue, e.g., lexical.
Activity 3. Translate the text using chat GPT and then edit the translation individually. After that, in pairs, compare the obtained results and present them in open class or in writing. Finally, submit the assignment either on Moodle or Padlet, or hand in a printed version to the teacher.
Activity 4. Translate the text individually. Ask ChatGPT to edit it. Analyse all the corrections subdividing them into the following categories: lexical, grammatical, syntactic, stylistic and pragmatic. Justify the necessity of each suggested correction. Submit your assignment on Moodle.
Based on the analysis of submitted tasks during the experiment and open class discussions with the students involved, we can infer that the offered activities helped students to identify typical mistakes made by Chat GPT faster, target weak points more effectively and improve collaborative skills when working in pairs and small groups.
Conclusions
The results of the carried our research show that in the digital era such technological advancements as Сhat GPT cannot be avoided. Moreover, they should be incorporated into the process of professional training of modern translators. The conducted pedagogical experiment demonstrated that, when integrated into translation courses through carefully designed activities, ChatGPT can enhance students’ translation competence, critical thinking, digital literacy, and post-editing skills. The results of this study demonstrates that despite its considerable potential, the tool still produces lexical, grammatical, stylistic, pragmatic, and context-related errors that require careful human evaluation and revision. These findings highlight that ChatGPT should be viewed not as a replacement for a human translator but as an educational tool that supports the development of professional competencies. Based on the results of the analysis of ChatGPT’s failures, a set of learning activities involving the tool and post-editing was designed to foster students’ ability to critically assess AI-generated translations, identify and correct errors, formulate effective prompts, and make informed translation decisions. The proposed activities demonstrate practical ways of integrating generative AI into translator training while maintaining high standards of translation quality and academic integrity.
Overall, the study suggests that the effective integration of ChatGPT into translator education requires a balanced pedagogical approach that combines AI-assisted translation with systematic post-editing, explicit guidance from lecturers, and carefully selected tasks aligned with students’ proficiency levels and learning outcomes.
While this study addresses crucial aspects, it acknowledges the incomplete exploration of the scientific problem, suggesting future research avenues, particularly in understanding intercultural interactions among specialists from diverse fields and developing pertinent methodologies.
Acknowledgements
The authors gratefully acknowledge the third-year students of the educational program “English-Ukrainian Translation” in Foreign Languages Department of Ternopil Volodymyr Hnatiuk National Pedagogical University (Ukraine) for taking part in this research.
NOTES
1. Calculate payroll deductions and contributions. About the deduction of Employment Insurance (EI) premiums. Government of Canada (2022, November 15). https://www.canada.ca/en/revenue-agency/services/tax/businesses/topics/payroll/payroll-deductions-contributions/employment-insurance-ei.html
2. Collins Dictionary. https://dictionary.cambridge.org/
3. European Master’s in Translation Competence Framework 2022. https://webgate.ec.europa.eu/circabc-ewpp/d/d/workspace/SpacesStore/f28e82d1-ac0d-4545-aa3f-dff2f0500fda/download
4. Ternopil Volodymyr Hnatiuk National Pedagogical University. (2015). Policy on the Use of Artificial Intelligence in Scientific, Technical, and Innovative Activities at Ternopil Volodymyr Hnatiuk National Pedagogical University. https://tnpu.edu.ua/about/public_inform/upload/2025/polityka_AI.pdf
REFERENCES
Atlas, S. (2023). ChatGPT for Higher Education and Professional Development: A Guide to Conversational AI. https://digitalcommons.uri.edu/cgi/viewcontent.cgi?article=1547&context=cba_facpubs.
Carmo, F. & Moorkens, J. (2020, October 26). Differentiating Editing, Post-Editing and Revision. In: Maarit Koponen, Brian Mossop, Isabelle S. Robert, Giovanna Scocchera (Eds.). Translation Revision and Post-editing. London. Routledge. https://doi.org/10.4324/9781003096962 https://www. researchgate. net/publication/343920013_Differentiating_Editing_Post-Editing_and_Revision.
Jiao, Wenxiang & Wang, Wenxuan & Huang, Jen-Tse & Wang, Xing & Shi, Shuming & Tu, Zhaopeng. (2023). Is ChatGPT a Good Translator? A Preliminary Study. https://www.researchgate.net/publication/367359399_Is_ChatGPT_A_Good_Translator_A_Preliminary_Study.
Karaban, V., Karaban, A. (2025). AI Era Classroom Beyond Post-Editing: Custom GPTs In Translator Training. The journal of V. N. Karazin Kharkiv National University. Series: Foreign Philology. Methods of Foreign Language Teaching, (102), 92 – 101. https://doi.org/10.26565/2786-5312-2025-102-11.
Karaban, V. & Karaban, A. (2024). AI-translated poetry: Ivan Franko’s poems in GPT-3.5-driven machine and human-produced translations. Forum for Linguistic Studies, 6(1) . https://doi.org/10.59400/fls.v6i1.1994.
Khatser, G. O. (2024). The Usage ofAI Tools for Teaching Translation. Vcheni zapysky TNU imeni V.I. Vernadskoho. Seriia: Filolohia. Zhurnalistyka [Scientific Notes of Taurida National V. I. Vernadsky University. Series: Philology. Journalism]. Publishing House “Helmenevtyka”. 35(74) № 4. Part 2. https://www.philol.vernadskyjournals.in.ua/journals/2024/4_2024/part_2/15.pdf.
Khoshafah, F. (2023). ChatGPT for Arabic-English Translation: Evaluating the Accuracy. Research Square. 10.21203/rs.3.rs-2814154/v2.
Kohnke, L. &Moorhouse, B. &Zou, D. (2023). ChatGPTforLanguageTeaching and Learning. RELC Journal, 54(3) . 10.1177/00336882231162868.
Larroyed, A. (2023). Redefining Patent Translation: The Influence of ChatGPT and the Urgency to Align Patent Language Regimes in Europe with Progress in Translation Technology. GRUR International, 72(11), 1009 – 1017. https://doi.org/10.1093/grurint/ikad099.
Orăsan, Constantin. (2023). ChatGPT for translators: a survey. Proceedings of the First Workshop on NLP Tools and Resources for Translation and Interpreting Applications. https://aclanthology. org/2023. nlp4tia-1.10.pdf .
Ostapovych, O. & Ostapovych, N. & Mazurenko, Y. (2023). ChatGPT in Training of Philologists and Translators. Challenges and Perspectives. Naukovi zapysky Natsionalnoho universytetu “Ostrozka akademiia”: seriia “Filolohia” [Scientific Notes of Ostroh Academy National University: Philology Series]. Ostroh: Publishing House NaUOA. 17(85). рр. 200 – 205. https://doi.org/10.25264/2519-2558-2023-17(85)-200-205.
Peng, Keqin & Ding, Liang & Qihuang, Zhong & Shen, Li & Liu, Xuebo & Zhang, Min & Ouyang, Yuanxin & Tao, Dacheng (2023). Towards Making the Most of ChatGPT for Machine Translation. Findings of the Association for Computational Linguistics: EMNLP, рр. 5622 – 5633. 10.18653/v1/2023.findings-emnlp.373.
Siu, Sai Cheong. (2023). ChatGPT and GPT-4 for Professional Translators: Exploring the Potential of Large Language Models in Translation. Available at SSRN. 10.2139/ssrn.4448091.