This article explores translation techniques within the context of dubbing, comparing results obtained through human translation, artificial intelligence (AI)-based approaches, and hybrid workflows. By analyzing semantic accuracy, stylistic adaptation, and the preservation of cultural specificity, the research identifies key differences between these methods, defining their strengths and weaknesses. Particular attention is paid to the balance between efficiency and translation quality, as well as the ethical implications of using AI in the audiovisual field. The findings contribute to determining prospects for the optimal combination of human expertise and technological capabilities to create high-quality and culturally relevant content.
Keywords: audiovisual translation, dubbing, AI translation, human translation, cultural sensitivity, translation ethics, translation technology.
Introduction
The way films and TV-shows are understood and enjoyed by people around the world is greatly influenced by the way they are translated, especially through dubbing. Figuring out the best way to do this is a challenging area because what makes an adequate translation is complex. Everyone's seen a foreign film where the dubbing was so far from perfect, the voices didn't match, or the jokes just weren't funny. This highlights the crucial issue of creating reliabletranslations, because audiovisual translation (AVT) is a constrained translation due to the external limitations of the language and communicative situation [8, p. 4] .
For a long time, human translators were the only ones doing this, carefully crafting versions that connected with new audiences. But now there are many different translation applications and this is not the only trend, as neural machine translation is becoming more popular [3, p. 264] . Artificial intelligence (AI) has already become a powerful tool in translation, offering the potential for faster and reliable results. This raises two fundamental issues: Can AI truly capture the artistry and nuances of human translation? What are the specific strengths and weaknesses of each approach? This is also a well-documented issue that has attracted much attention.
The purpose of this study is to answer these questions by comparing three different ways to do dubbing: using only human translators, using only AI, and using a mix of the two. Each method was examined in terms of important criteria: how accurate it is, how natural it sounds, and how well it fits the new culture. Furthermore, the specific translation techniques used in each method were analyzed, noting what works well and what does not.
By carefully comparing these three approaches, this study seeks to provide valuable insights into the potential applications and limitations of AI in movie translation. The tasks of the study strive to fill the gap and aims not only to promote a better understanding of the adequate audiovisual translation methods in a globalized world, making it more enjoyable and meaningful for audiences, but also address an issue of growing interest within the field.
Literature Review
Now in the digital age, artificial intelligence has become prevalent in every aspect of our lives and translation is no exception. In addition, many people are using machine translation, which is now mostly based on artificial intelligence, in order to satisfy increasingly globalized and immediate consumption that aims to cover all possible local markets and respond to the demands of more and more consumers [1, p.6].
Nowadays, many people do not understand how to respond to new technologies in the form of AI, and MT in particular, which naturally causes concern among professionals and audiences in the field of AVT, but putting on a blindfold and acting as if the technology does not exist may not seem like the best approach to solving the problem [2, p. 162].
On the contrary, the protection of labor rights, the improvement of the working conditions of practitioners and the observance of certain ethical values are necessary, but they should always be approached from a reasonable and informed position. Moreover, there are many translators who consider that AVT is an interlanguage adaptation due to the fact that subtitling and dubbing offend traditional conceptions of an important category such as equivalence [7, p. 375].
Furthermore, the translation of audiovisual content requires a level of interpretation and creativity, or hermeneutics, which, for the time being, machines do not seem to be able to deliver [1, p. 11]. Creativity is a variety of different structures and styles of language that humans can do, but not a machine.
Methodology
The current study involved watching and analyzing two films and one TV series to identify an adequate translation appropriate to the cultural context. Moreover, this study focuses on various translation techniques including dubbing, human translation and artificial intelligence translation. The material was investigated through systematic observation and comparative analysis.
The materials are two films Tetris (2023) , Twisters (2024) and one TV series The Crown (2016–2023). A total of three dialogue extracts that have difficulties with the translation and cultural context. These exacts were analyzed in terms of comparative analysis which included quality assessment of translation like accuracy, cultural context and contextual meaning.
The study consisted of several steps. Firstly, choosing original extracts from English films. Secondly, searching for official translations from dubbed films made by professional translators. Thirdly, the extracts translated by the authors of the article. Finally, the machine translation was received using ChatGPT .
The method was based on subjective evaluations of translation and dubbing. The study applied the techniques of scientific observation and comparative analysis, which allowed for systematic data processing and objective evaluation of the results.
Contextual factors, such as tone of voice, cultural adaptations, and adherence to language norms, were carefully considered. Each comparison was specifically tailored to the particular media product, to minimize the influence of genre-specific characteristics on the results. Observations were conducted using strict and repeatable procedures, ensuring the reliability of the findings.
This research uses existing studies on comparing translations in audiovisual materials, including work by experts in dubbing and machine translation. Combining observation and experiment provided both theoretical knowledge and practical data.
Therefore, the methods used help ensure the accuracy of the results and allow for an objective understanding of the differences between dubbing, human translation, and AI translation. However, judging how natural a translation sounds is subjective, so we will use independent experts to minimize potential biases.
Results
This chapter presents a comparative analysis of three translation modalities — Artificial Intelligence (AI) translation, professional dubbing translation, and the author’s translation — across three distinct film extracts. The selected films, The Crown , Tetris , and Twisters , represent diverse genres and linguistic challenges, offering a rich landscape for exploring the strengths and weaknesses of each approach.
To facilitate a structured evaluation, the three translations (dubbed, human, AI) were subjected to a rigorous comparative analysis based on three key criteria:
– Accuracy: The extent to which the translation faithfully conveys the literal meaning of the source text.
– Cultural Context: The effectiveness of the translation in adapting cultural references and idiomatic expressions for the target audience.
– Сontextual Meaning: The degree to which the translation maintains the overall tone, intent, and impact of the original dialogue.
The analysis focuses on identifying key differences in semantic accuracy, stylistic choices, adaptation strategies, and overall effectiveness in conveying the intended meaning and emotional impact of the original source material.
Each film extract underwent individual analysis to provide a detailed comparison of how each translation modality addressed specific instances of dialogue, cultural references, and idiomatic expressions. The ensuing findings will illuminate the current capabilities of AI translation, the techniques employed in professional dubbing, and the nuanced approach characteristic of human translation. Ultimately, this analysis contributes to a more comprehensive understanding of the evolving translation landscape.
The following table offers a concise overview of the performance of each translation method concerning the defined criteria, thereby enabling a direct comparison of their relative strengths and limitations across the selected film extracts. The subsequent sections will present a more in-depth discussion of these findings, supported by illustrative examples and a nuanced assessment of the effectiveness of each translation in the Table 1 .
Table 1
|
The name of the film |
The original text of the film |
Dubbing |
ChatGPT |
Translation options for the authors of the article |
|
The Crown, 2016–2023 |
“- You don’t have to make such a fuss of him. — He was nervous. He signed in the wrong place. — Maybe because she was standing behind him? — Smothering him? — Mothering . — Smothering .” |
- Не надо было стоять у него над душой. — Он нервничал, не там расписался. — Может, потому что ты стояла за ним, смущала его? —Я успокаивала . — Смущала . |
- Не стоит так с ним носиться. — Он нервничал. Подписался не там, где нужно. — Может, потому что она стояла у него за спиной? — Задушила его своей опекой? — Материнская забота . — Именно, удушающая опека . |
- Не стоит его так — Он нервничал, не там расписался. — Может, потому что ты водишься с ним? — Я забочусь о нём. — Ты водишься с ним. |
|
Tetris, 2023 |
“-You're in need of succour ? — Excuse me? — Succour , noun, assistance in times of stress. — Synonyms with help. Sasha, Translation Services. Esteemed to meet you.” |
- Чем могу подсобить ? — Простите? — Подсобить , глагол, помочь в трудную минуту, прийти на помощь. — Саша, переводчица. Польщена знакомством. |
- Вам требуется поддержка ? — Простите? — Поддержка , существительное, помощь во время трудностей. Синоним — помощь. Саша, из Службы Переводов. Рад знакомству. |
- Вам нужна помощь ? — Что вы имеете в виду? — Помощь , существительное, означает помочь в трудную минуту, от глагола помогать. Я — Саша, переводчик, приятно познакомиться. |
|
Twisters, 2024 |
“- Please, stop encouraging her. — Keep your clothes on , Kate. Let your mom finish.” |
- Хватит её поощрять. — Из штанов не выскакивай . Дай ей закончить. |
- Прошу, перестань её подначивать. — Кейт, не раздевайся раньше времени . Дай маме закончить. |
- Хватит ей потакать. — Кейт, успокойся , дай маме закончить. |
The table demonstrates a multifaceted comparison of the three translation modalities — Artificial Intelligence (AI) translation, professional dubbing translation, and human-generated translation — across three distinct film extracts. Each method exhibits distinct strengths and weaknesses in capturing the nuances of the original dialogue. These variations are evident not only in the accuracy of the translations but also in their ability to adapt to cultural contexts and preserve the contextual meaning.
To further clarify the methodologies used, the human translation involved a collaborative effort, with the authors working together to refine and agree upon the most suitable rendering of each extract. This process included multiple iterations, incorporating discussions on the specific connotations and subtext of the original lines.
For a more detailed view, please refer to the table above, which provides a side-by-side comparison of the original text and the three translation versions.
The AI-generated translations, while often accurate on a literal level, sometimes struggled to capture the idiomatic expressions and cultural subtleties present in the source material. In contrast, the dubbed translations frequently demonstrated a strong ability to adapt to cultural contexts, occasionally at the expense of strict literal accuracy.
Specifically, the human translations in The Crown extract effectively captured the underlying tension and relational dynamics, with particular attention paid to the wordplay of « smothering» and « mothering». The human translators carefully considered the suggestive meaning and the contextual implications, providing a more nuanced alternative than the AI or dubbed versions.
In the Tetris extract, the challenge lay in preserving the original meaning of « succour », a word chosen for its specific register and the fact that it's a homophone. The human translation prioritized maintaining this original meaning, recognizing its importance in the context of the scene, whereas the other translations varied in their success. In addition, the option of translating the word ‘translator’ chose the masculine option of this word in Russian.
TheTwisters extract presented a different challenge: the idiomatic expression « keep your clothes on ». In this case, the human translation stood out by providing the most accurate and contextually appropriate rendering of the idiom. The AI and dubbed translations failed to capture the idiomatic meaning, offering more literal interpretations.
Overall, the human translations tended to strike a balance between accuracy and cultural sensitivity, demonstrating a more nuanced understanding of the source text. This highlights the value of human translators in navigating the complexities of language and context.
The analysis relied on a combination of qualitative assessment and comparative analysis, focusing on identifying key differences in semantic accuracy, stylistic choices, and adaptation strategies.
While the subjective nature of evaluating translation, quality presents an inherent challenge, the use of multiple evaluators and clearly defined criteria helped to mitigate potential biases. Moreover, the focus on comparative analysis allowed for a systematic identification of patterns and trends across the different translation modalities.
The implications of these findings extend to the broader field of audiovisual translation, informing discussions on the role of AI in the translation workflow and highlighting the continued importance of human expertise in ensuring high-quality and culturally sensitive translations.
Conclusion
Our investigation comparing human, AI, and human-assisted dubbing translation reveals a rich picture of the evolving landscape of audiovisual translation. By analyzing how each handle meaning, style, and cultural nuances, we’ve clarified their unique strengths and limitations.
The analysis revealed a consistent strength in human translation's ability to navigate complex linguistic and cultural nuances. Human translators demonstrated a superior capacity for crafting translations that resonate authentically with the target audience, often achieving a more effective equilibrium between accuracy and cultural relevance than purely AI-driven approaches. This capacity hinges on a profound understanding of cultural context and audience expectations, skills currently surpassing the capabilities of AI.
However, AI-powered translation is undeniably advancing. While AI can't yet fully replicate human nuance, its efficiency and scalability offer great potential to boost the translation workflow. Smart integration of AI, for example in initial drafts and managing terminology, could increase productivity and cut costs.
However, ethical considerations are paramount. Relying on AI raises questions of authorship, creativity, and potential algorithmic bias. Future research must explore mitigating these risks, ensuring AI is used responsibly and ethically, with human oversight guaranteeing fairness and accuracy, especially in culturally sensitive areas.
Acknowledging the subjectivity of translation evaluation as a study limitation suggests fruitful avenues for future inquiry. Longitudinal studies are crucial to understand AI’s long-term impact on translation quality and the evolving role of human translators in a tech-driven environment. Further research should also compare the effectiveness of different AI training datasets and algorithms, focusing on boosting cultural awareness.
The future of audiovisual translation hinges on humans and machines working synergistically. By leveraging AI’s efficiency and scalability while valuing human translators’ cultural expertise and nuanced interpretation, we can strive for translations that are not only accurate but also captivating, culturally fitting, and ethically sound. The real challenge lies in harnessing technology while championing linguistic diversity, cultural sensitivity, and human ingenuity in the global exchange of audiovisual content.
References:
- Celia, R. P. caminos convergentes en traducción audiovisual, traducción automática y posedición / R. P. Celia. — Текст: непосредственный // La traducción audiovisual a través de la traducción automática y la posedición Prácticas actuales y futuras. — 2023. — № 343. — С. 1–14.
- Frederic, Chaume Audiovisual Translation: Dubbing / Chaume Frederic. — 1. — London: Routledge, 2012. — 228 c. — Текст: непосредственный.
- Hilary, Glasman-Deal Science Reserch Writing For Non-Native Speakers of English / Glasman-Deal Hilary. — London: Imperial College Press, 2010. — 245 c. — Текст: непосредственный.
- Jorge, L. R. Hermēneus, 20 (2018): págs. 257–294 ISSN: 2530–609X Aspects of human translation: the current situation and an emerging trend / L. R. Jorge. — Текст: непосредственный // Hermēneus. — Málaga:, 2017. — С. 257–294.
- Laura, Mejías-Climent Beyond the black mirror effect: the impact of machine translation in the audiovisual translation environment / Mejías-Climent Laura, de,Los,Reyes Julio. — Текст: непосредственный // Linguistica Antverpiensia New Series. — 2023. — № 20. — С. 19.
- А. А. Альварес Роль аудиовизуального перевода в обеспечении межкультурного и межъязыкового взаимодействия / А. А. Альварес — Текст: непосредственный // Вестник МГЛУ. — 2018. — № 5. — С. 7.
- Козуляев, А. В. Аудиовизуальный полисемантический перевод как особая форма переводческой деятельности и особенности обучения данному виду перевода / А. В. Козуляев. — Текст: непосредственный // Царскосельские чтения. — 2013. — № XVII. — С. 374–381.
- Козуляев, А. В. Обучение динамически эквивалентному переводу аудиовизуальных произведений: опыт разработки и освоения инновационных методик в рамках школы аудиовизуального перевода / А. В. Козуляев. — Текст: непосредственный // Вестник ПНИПУ. — 215. — № 3. — С. 22.
- Комарова З. И. Методология, метод, методика и технология научных исследований в лингвистике / З. И. Комарова. — Екатеринбург: УрФУ, 2012. — 818 c. — Текст: непосредственный.
- Новые Аудиовизуальные Технологии / Грановская О. В., Дуков Е. В., Иоскевич Я. Б. [и др.]. — Москва: Едиториал УРСС, 2005. — 479 c. — Текст: непосредственный.
