日本語・ベトナム語翻訳 – 翻訳者による訳文における質と機械翻訳による訳文における質の比較 = biên dịch nhật việt – so sánh chất lượng bản dịch do người dịch và chất lượng bản dịch do máy dịch
はじめに
研究課題を選択する理由
In modern life, rapid societal changes and technological advancements have led to numerous scientific achievements that significantly influence our daily experiences These breakthroughs have transformed industries, improved healthcare, and enhanced communication worldwide Embracing these innovations is essential to stay competitive and adapt to the evolving landscape Understanding the importance of scientific progress helps individuals and organizations leverage new technologies for growth and development Staying informed about recent scientific discoveries ensures we maximize their benefits and address potential challenges effectively.
As a Vietnamese-Japanese translator, I frequently use machine and automated translation applications; however, I have realized that the quality of human translator's work often surpasses that of these tools This discrepancy highlights the gap in translation quality between professional translators and machine translation applications Bridging this gap could reduce the workload for human translators and enable machine translation tools to become valuable assets, increasing translation speed Therefore, I am interested in researching methods to eliminate this quality gap and improve the overall effectiveness of translation technology.
Most of the existing research has focused on similar themes related to translation, but the majority have studied machine translation and human translation separately Notably, there is a lack of studies comparing and systematizing machine-translated texts with human-translated texts, highlighting a significant gap in comparative translation research.
研究の目的
Hiện nay, các dịch giả cần sử dụng công cụ dịch thuật để hỗ trợ quá trình dịch Công cụ dịch thuật AI đóng vai trò quan trọng trong việc nâng cao hiệu quả và độ chính xác của bản dịch, giúp quá trình dịch nhanh hơn và phù hợp với các yêu cầu về chất lượng nội dung.
Chúng tôi đã cải thiện đáng kể quá trình dịch thuật, giúp giảm thiểu công sức và tăng hiệu quả công việc Trong tiếng Anh, các công cụ dịch và trí tuệ nhân tạo hoạt động rất tốt, mang lại khả năng dịch chính xác và nhanh chóng Tuy nhiên, đối với các cặp ngôn ngữ như tiếng Nhật – tiếng Việt và tiếng Việt – tiếng Nhật, còn nhiều hạn chế cần khắc phục để nâng cao chất lượng dịch thuật.
This study aims to compare human translation by professional translators with machine translation to identify key differences It emphasizes the importance of accurate grammar and vocabulary to minimize errors The findings highlight the distinct characteristics of each translation method, providing insights for professionals working as translators and editors to improve quality and consistency in their work.
表現、文脈などの誤りを体系化することである。翻訳者がその文法、語彙、表現などの誤りの体系を利用して、より良い日越翻訳ができるような目的である。
研究の質問
上記の理由と目的の上に、次の質問に答えるために調査を実施した。
1 翻訳ツールは本当に効果があるか。
2 翻訳者による翻訳と機械翻訳による翻訳の違いは何か。
This study investigates errors in machine translation by analyzing and categorizing common mistakes in translated texts Based on these findings, we developed essential guidelines for translators and editors to improve translation accuracy and quality.
研究の範囲
This study investigates the extent of translation tool usage among Japanese language translators and workers, gathering insights from students and professionals both within and outside Hanoi University and the University of Foreign Languages.
This study compares translations generated by translation tools with those produced by experienced human translators The two versions are analyzed and grouped based on errors related to grammar, vocabulary, and context To ensure objective evaluation, a selection of documents was randomly chosen for the comparison The research utilizes specific works and materials to facilitate an accurate assessment of translation quality.
1 『羊と鋼の森』（宮下奈都 著）作品と文学出版社の翻訳者Nguyễn Hải Hàによる
『Người sửa đàn dương cầm』版（1ページ― 10ページ）、
2 『夏の庭』（湯本香樹実 著）作品と文学出版社の翻訳者Nguyễn Thanh Hàによる
『Totto chan』版（126ページ―134ページ）、
Trong phần này, bài viết đề cập đến tác phẩm 『窓際のトットちゃん』 của tác giả黒柳徹子 và bản dịch 『Totto chan – Cô bé ngồi bên cửa sổ』 do dịch giả Trương Thùy Lan thực hiện, từ trang 184 đến 191 Bài viết nhấn mạnh tầm quan trọng của nội dung câu chuyện và vai trò của dịch giả trong việc truyền tải giá trị văn học từ tiếng Nhật sang tiếng Việt Đây là một phần quan trọng giúp độc giả hiểu rõ hơn về quá trình dịch thuật cũng như ý nghĩa sâu sắc của tác phẩm đối với văn học thiếu nhi và văn hóa Nhật Bản.
This study uses three different works translated by three separate translators to enable an objective comparison of translation quality and style.
Different translations by three different translators result in varied interpretations, highlighting that comparisons with machine translation should not be limited to just a single human translation.
The translation software used for comparison with the author's translated text is Google Translate.
研究の構成
This study is structured into five chapters Chapter 1 provides an overview of the research paper, outlining its main focus Chapter 2 reviews relevant theories and prior research, including fundamental translation theories, machine translation and automatic translation, types and advantages/disadvantages of translation tools, theories related to automatic translation logic, and overall translation theories It also compares human translations by experienced translators with machine-generated translations Chapter 3 details the research methods employed, such as research background, data collection and analysis techniques, target subjects, and presents tables listing errors between machine-translated texts and human translations Chapter 4 discusses the findings from investigating the use of machine translation by translators and editors, highlighting key content and cautionary notes Finally, Chapter 5 offers conclusions that summarize key results, address research limitations, and suggest future research directions, ensuring adherence to SEO best practices.
研究テーマに関する理論
翻訳と機械翻訳に関する理論
2.1.1 翻訳の基本理論
According to Chapter 2, "Basic Principles of Translation," on page 2 of the Smart Korean Study website, translation is understood as "transformation," "interpretation," or "explanation" in Chinese Dictionary definitions describe translation as the "rendering of text from one language to another" or the "conversion of speech or writing from one country's language to another."
例： 日本語 ➞ Tiếng Nhật
マイさんは日本語を勉強している ➞ Mai đang học tiếng Nhật
According to this study, translation is defined as the process of logically converting source language documents into a target language while maintaining lexical and stylistic norms The appropriateness of this translation process is determined by the translator's ability, and it is influenced by factors such as the translation process, psychological factors, the translator’s experience, and situational conditions.
本研究はまた、翻訳を定義するときに与えられる焦点と方向性を把握するために、 エティエンヌ・ドレが次のように指定された5つの原則を考慮することを示した。
（1）翻訳者は、物事を明確にし、オリジナルの美しさと意味を理解するために柔軟でなけ ればならない。
A translator must possess comprehensive knowledge of both the source and target languages to ensure accurate and high-quality translation Understanding the nuances, grammar, and cultural contexts of each language is essential for delivering precise and natural results This expertise enables the translator to accurately interpret the original content and convey its meaning effectively in the target language.
（3）翻訳者は単語から単語への訳文を避ける必要がある。
（4）翻訳者は、日常の言葉で一般的な単語を使用する必要がある。
Translators must carefully select and use words to create correct intonation, ensuring that the intended meaning and natural flow of speech are accurately conveyed Proper word choice is essential for achieving appropriate pronunciation and delivering a coherent, effective translation.
2.1.2 機械翻訳と自動翻訳の基本理論
「機械翻訳」とは何か?
Machine translation refers to the process of using computers to perform translations instead of humans This method has been around for a long time, starting with rule-based machine translation, which translates based on pre-registered linguistic rules (NTT Communications)
Nguyen Truong Long（2019）によると、自動翻訳とも呼ばれる機械翻訳は、ソフトウェ
Computational linguistics, a subfield dedicated to the study of language usage, focuses on developing software that translates text or speech from one natural language to another This interdisciplinary field combines ideas and techniques from linguistics, computer science, statistical probability, and artificial intelligence The primary goal of machine translation is to create systems that enable accurate and natural translations between human languages.
The concept of machine translation began to emerge in the 17th century However, it wasn't until the 20th century that concrete ideas and developments regarding machine translation started to be recognized and explored.
機械翻訳と自動翻訳の種類
機械翻訳には、次の3つの種類がある。
• ルールベース機械翻訳
Rule-based machine translation is a system that generates translations by applying grammatical rules derived from accumulated translation data, vocabulary, and sample sentences The process begins with a grammatical analysis of the source text’s structure and meaning, followed by translating this analysis into the target language This approach ensures that the translation adheres to linguistic rules, providing accurate and coherent results.
• 統計ベース機械翻訳(SMT)
Statistical machine translation learns translation models from large stored translation data, enabling accurate and effective translations This approach generally produces higher-quality translations compared to rule-based machine translation systems.
• ニューラル機械翻訳(NMT)
Neural machine translation utilizes neural networks that mimic the structure of human brain neurons This technology is based on systems that replicate how the human brain processes and remembers language, enabling more accurate and context-aware machine translation.
Among the three methods, natural translation is highly accurate and can capture even subtle nuances, ensuring the translated content remains faithful to the original meaning.
現在の4.0の時代では、多くの種類の自動翻訳および機械翻訳ツールを見つけて使用することは難しくない。以下の日本語のテキストを翻訳できる翻訳ツールの種類を検索して見つけた。
• まず、次のような現在の無料のオンライン翻訳ツール製品について。
Google翻訳、Microsoft Translator、Mazii、Weblio、Linguee、Tangorin、Jdict、Kantanなど
Professional translation software, also known as Computer-Assisted Translation (CAT) tools, are essential for meeting high-quality translation needs Popular CAT tools include SDL Trados, Wordfast, and MemoQ, which enhance translation efficiency and consistency These programs support professional translators by providing advanced features such as translation memory, terminology management, and project collaboration, ensuring accurate and timely delivery of translation projects Incorporating CAT software into your workflow can significantly improve productivity and maintain high standards in multilingual content creation.
Heartsome、DéjàVu（DV）などの世界中の翻訳会社によって提供されている。
（Viettranslation、2018）。これらの翻訳ソフトウェアのほとんどは、大学生にとっ
先行研究
This article references numerous prior studies on general translation, with a particular focus on comparing translated texts to analyze differences and improve translation quality.
• 自動翻訳ツールの使用の度合に関する調査の研究
Trong nghiên cứu này, chúng tôi đã sử dụng một khảo sát tương tự như trong bài viết của Công Huyền Tôn Nữ Ý Nhiệm (2016) về "Dịch tự động và kỹ năng viết của sinh viên", tuy nhiên, đã có một sự điều chỉnh ở một câu hỏi trong khảo sát để phù hợp hơn với mục tiêu nghiên cứu.
講師のCong Huyen Ton Nu Y Nhiem（2016）による研究論文は、フエ大学のフランス語外国
Sinh viên trong khoa ngôn ngữ thường xuyên sử dụng công cụ dịch tự động, gây ảnh hưởng tiêu cực đến quá trình học tập của họ Các nghiên cứu đã đề xuất các giải pháp tạm thời để các giảng viên hạn chế tác động xấu này và bảo vệ chất lượng học tập của sinh viên.
• 翻訳における翻訳ソフトウェアの適用に関する研究
Nguyễn Ninh Bắc（2014）の調査結果によると、同じ英語のテキストを2つの自動翻
When comparing the translation applications Vdict and Google Translate, it becomes evident that Google Translate offers superior performance and accuracy.
Vdict's products are more accurate and natural compared to other options However, they primarily focus on providing basic information, highlighting their strengths in delivering precise language tools while acknowledging limitations in more advanced data.
If the sole purpose is translation, you can fully rely on Google Translate with minimal need for a professional translator's involvement.
In addition to comparing the translation quality of two free translation tools, he also emphasizes the importance of understanding their specific features and limitations.
Phần mềm hỗ trợ dịch thuật (CAT) được hiểu là phần mềm máy tính giúp tăng tốc quá trình dịch và cải thiện sự nhất quán của thuật ngữ nhờ nguyên lý hoạt động đặc biệt Sử dụng đúng cách phần mềm CAT mang lại hiệu quả vượt trội, giúp tăng năng suất dịch hơn 30%, duy trì sự nhất quán của thuật ngữ, xây dựng từ điển thuật ngữ cá nhân và hỗ trợ làm việc nhóm hiệu quả hơn.
彼はまた、翻訳市場で最も人気のあるソフトウェアはSDL TradosとWordfastであ
り、OmegaT.SDLTradosは世界で最も使用されていると指摘した。上記の翻訳ソフトウェア
のコストも大学生に比べてかなり高いである。
This study reveals that WF Software is included in the training program at Hanoi National University of Foreign Studies, designed specifically for students studying technology and translation.
Nguyễn Ninh Bắcはまた、次のように述べている。「CATソフトウェアを使用する際
Trong lĩnh vực dịch thuật, cần chú ý rằng phần mềm dịch không phải là công cụ thay thế hoàn toàn cho người dịch giỏi, mà còn có thể giúp cải thiện chất lượng bản dịch kém trở nên tốt hơn, thậm chí như một phép màu Chất lượng dịch thuật chủ yếu phụ thuộc vào năng lực chuyên môn của người dịch, chứ không phải vào phần mềm sử dụng Công cụ CAT hỗ trợ tăng tốc độ dịch và duy trì tính nhất quán trong thuật ngữ cho các loại tài liệu cụ thể, đặc biệt là các tài liệu có nhiều cụm từ và thuật ngữ lặp lại như tài chính, môi trường hay công nghệ Tuy nhiên, trong dịch phẩm văn học hay tiểu thuyết, phần mềm CAT gần như không mang lại hiệu quả đáng kể.
This article provides an overview of information technology applications in translation work by Nguyen Ninh Bak, focusing on current translation software, including free and paid options It discusses the advantages, disadvantages, and costs associated with professional computer-assisted translation and emphasizes that paid software often offers higher translation quality However, as a research paper on the application of information technology in translation, it lacks extensive comparative data on different translation tools and software.
• 翻訳品質の評価に関する研究
Nguyễn Thanh Tùng（2009）、ホーチミン市教育大学英語学部は、最終試験のデータに
基づいて、英語学部の学生のクラスの翻訳能力と質を評価するための調査を実施した。語
彙、基本レベルの文法、学生のクラスのレベル、および学生のスコアの統計の観点から、 英語の翻訳における質を調査および批評する。（p.p 27-32）。
Nguyễn Thanh Tùng là một sinh viên đang nỗ lực phát triển các kỹ năng về từ vựng, ngữ pháp và dịch thuật, góp phần nâng cao trình độ ngôn ngữ của mình Anh ấy tập trung vào việc mở rộng vốn từ và cải thiện khả năng sử dụng ngữ pháp chính xác để tự tin hơn trong giao tiếp và học tập Ngoài ra, Nguyễn Thanh Tùng còn chú trọng vào việc thực hành dịch thuật để nâng cao khả năng chuyển đổi ngôn ngữ một cách chính xác và hiệu quả Việc nâng cao các kỹ năng này giúp anh ấy xây dựng nền tảng vững chắc cho ngành dịch thuật và ngôn ngữ học trong tương lai.
研究の方法
研究の背景
In the era of Technology 4.0, which involves diverse development of translation tools, students have access to various translation technologies However, the distinction between proper technology application and misuse is often unclear Additionally, machine translation cannot yet achieve 100% accuracy, highlighting the ongoing need for human oversight and expertise in translation.
There is a growing demand for document and book translations between Japanese and Vietnamese, leading to an increase in available works in the market Technology 4.0 has been widely integrated into various fields, including book translation, enhancing efficiency and accuracy Translators now utilize both their expertise and translation tools to support their workflow However, a key concern is whether translators should rely entirely on machine translation without further proofreading when using translation tools.
This study highlights the differences between machine translation and human translation It also explores how machine translation can be effectively used to translate documents and books Additionally, the research provides useful notes for professionals in the translation industry and editors of books, emphasizing best practices when utilizing machine translation tools.
研究の手段
To assess users' proficiency with translation tools, a survey utilizing both closed and open-ended questions was conducted The closed questions included multiple-choice options to gather quantitative data, while the open-ended questions provided respondents with the opportunity to share detailed insights and personal experiences This comprehensive approach ensures a thorough understanding of users' translation tool usage and skill levels.
In the translation comparison section, researchers manually perform comparisons using comparison tables due to limited conditions for automated translation analysis Additionally, there are no machines available to measure translation accuracy, and each translation depends on various aspects from the translator’s perspective, making absolute accuracy unattainable Therefore, the evaluation is only relative rather than absolute Consequently, I consulted with the instructor during the translation comparison part to ensure accuracy and reliability.
研究の対象
The research phase involves individuals responsible for Japanese translation work, including students from Hanoi National University and Foreign Language University, as well as professionals and workers involved in the project, ensuring diverse participation from both inside and outside the university.
In the translation comparison section, experienced translators' versions are used to compare with machine-generated translations, ensuring accurate assessment of translation quality.
Professionally translated documents authored or published by experienced translators ensure high-quality, accurate content These translations adhere to SEO best practices, enhancing visibility and reach Utilizing expert translation services guarantees that your content communicates effectively across languages, making it ideal for publication or online posting.
This study utilizes documents translated by experienced translators who have a background in book translation These translations serve as the basis for our analysis, providing insights into translation quality and techniques By examining their work, we aim to understand effective strategies in literary translation and improve future translation practices The research emphasizes the importance of professional expertise in maintaining the integrity and nuance of original texts during the translation process.
This article discusses books published by Vietnamese and Japanese literary publishers, with Japanese texts translated and available on Japanese websites, including the Vietnamese Embassy The materials used in this research are carefully selected, with a focus on objective criticism, and the number of investigation documents is determined through random sampling to ensure unbiased analysis.
1 『羊と鋼の森』（宮下奈都 著）作品と文学出版社の翻訳者Nguyễn Hải Hàによる
『Người sửa đàn dương cầm』版（1ページ― 10ページ）、
2 『夏の庭』（湯本香樹実 著）作品と文学出版社の翻訳者Nguyễn Thanh Hàによる
『Khu vườn mùa hạ』版（126ページ―134ページ）、
Trong tác phẩm 『窓際のトットちゃん』 của nhà văn 黒柳徹子, câu chuyện kể về cô bé Trò Truyện sống bên cửa sổ, thể hiện nét đẹp trong sáng và trí tuệ của trẻ thơ Phiên bản翻訳 của nhà ngôn ngữ học Trương Thùy Lan mang tên 『Totto chan – Cô bé ngồi bên cửa sổ』 đã chuyển thể thành công nội dung này từ trang 184 đến 191, giúp độc giả Việt Nam dễ dàng tiếp cận và hiểu rõ hơn về câu chuyện cảm động này Bản dịch không chỉ giữ nguyên tinh thần của tác phẩm gốc mà còn tối ưu hóa cho các tiêu chí SEO, giúp nâng cao khả năng tìm kiếm và tiếp cận thông tin của người đọc.
『羊と鋼の森』の翻訳について
『Người sửa đàn dương cầm』というタイトルの作品の翻訳は、現在ハノイ国家大学・外
Bài viết này được dịch bởi giảng viên Ngôn ngữ Nhật và Văn hóa Nhật Bản tại Đại học Ngôn ngữ, dịch giả Nguyễn Hải Hà Trong nghiên cứu này, chúng tôi sử dụng phiên bản do nhà xuất bản văn học phát hành vào năm 2018 để đảm bảo tính chính xác và cập nhật của nội dung.
『夏の庭』の翻訳について
Kaoruki Yumoto was born in Tokyo in 1959 and studied composition at Tokyo University of Music During her studies, she wrote lyrics for operas and had her work performed on radio and television Her debut novel, "Marigarden" (Summer Garden), published in 1992, quickly gained fame both domestically and internationally The novel was nominated for the German Youth Literature Award in 1996 and won the prestigious Boston Globe-Horn Book Award for Children's Literature in 1997 Currently residing in Tokyo, Kaoruki Yumoto continues to write.
『Khu vườn mùa hạ』というタイトルの作品のベトナム語訳文は、2010年に発行され、
The document was reissued in 2017, utilizing the Vietnamese edition originally published in 2010.
『窓際のトットちゃん』の翻訳について
"Totto-chan by the window has been a beloved bedside book for children worldwide for over 30 years Since its first publication in 1981, it has become a bestseller not only in Japan but across the globe As of 2001, the total number of copies sold in Japan reached 9.3 million, making it one of the best-selling books in the country's publishing history."
The English version of "Totto-chan" was published in the United States, leading the New York Times to immediately publish two full-page feature articles, an honor rarely bestowed upon other works Additionally, the book has been translated into 33 different languages, reflecting its global popularity and cultural significance.
本調査では、2011年に文学出版社から発行されたTrương Thuỳ Lanによる『Totto chan –
Trong tác phẩm "Cô bé ngồi bên cửa sổ", nhân vật chính là một cô bé trẻ tuổi, thể hiện qua cảnh cô ngồi bên cửa sổ Khi tác phẩm được dịch sang tiếng Việt, cô bé này vẫn còn rất nhỏ tuổi, mang đến cảm giác trong sáng và thuần khiết Nội dung tác phẩm nhấn mạnh vẻ đẹp của những khoảnh khắc đơn giản nhưng đầy ý nghĩa trong cuộc sống của tuổi trẻ Bằng cách tập trung vào hình ảnh cô bé bên cửa sổ, bài viết tạo nên một không khí lắng đọng, gợi nhớ về sự trong sáng của tuổi thơ và những cảm xúc chân thật.
しかし、翻訳は出版され、読者から好評を博している。
機械翻訳と自動翻訳によって翻訳された同様の書類
This study examines both the referenced materials and documents translated using Google Translate The choice of Google Translate was based on the high popularity of this product and its relevance to the reasons for selecting this task The translated content via Google Translate was produced in March 2021.
データ収集の方法
The research involved collecting primary data through a survey using questionnaires designed specifically for students of the Faculty of Japanese Language and Culture at Hanoi National University and Foreign Language University Most questions were closed-ended with predefined response options, while some included open-ended options to allow respondents to share additional explanations The survey was distributed as links via social networking platforms, such as Facebook, enabling participants to complete the online questionnaire conveniently.
In the translation comparison section, the study utilizes secondary data collection methods, referencing various existing data sources such as research papers, articles, essays, internet posts, instructional materials from instructors, and published literary works This approach allows for a comprehensive analysis by leveraging already available data across multiple platforms.
データ分析の方法
The survey section involves conducting research using statistical analysis methods on the collected quantitative data.
In the translation comparison section, we utilize content analysis of qualitative data, focusing on a comparative analysis between machine translation outputs and human translator results to evaluate translation quality.
Researchers collect Japanese works and texts, translate them, and publish them for public access They then analyze and categorize the errors and limitations of machine translations in comparison to human translations These machine translation errors are classified based on aspects such as vocabulary, grammar, and context, providing insights into their strengths and weaknesses.
To ensure an objective comparison and accurate commentary, the comparison table is analyzed by expert teachers and researchers, who provide insights based on the theories outlined in Chapter 2, ensuring a thorough and credible evaluation.