Speaker: Kazumasa Aoyama
Session Title: Common Mistakes in English-to-Japanese Scientific Translation, presented by Kazumasa Aoyama
Review Author: Katherine Michiko Hall
In this session on the first day of ATA66, Kazumasa Aoyama shared insights on common mistakes in English-to-Japanese scientific translation, the types of common mistakes, and survey results on how translators prevent and remedy these mistakes in their work. Mr. Aoyama’s presentation came from data gathered from a survey given to JLD members as well as his own research.
The first main survey question was about the types of translation mistakes that translators notice in their work: typos, omissions/additions, terminology errors, mistranslations, inconsistencies, style issues, format issues, grammar issues, punctuation errors, or “other.” The results of the survey showed that typos were most common, then mistranslations, then terminology errors, though Mr. Aoyama raised a good question about whether typos are the most common mistake because they are the easiest to notice.
Next was an analysis of English-to-Japanese patent translation errors, gathered from 150 samples from November 2024 to October 2025. All translations were human translations, not AI. The average number of words for each translation was 3,361 words; a total of 1,252 errors were found among all 150 translations, with an average error count of 4.97 per 2,000 words. The study found that of the errors identified, the most common involved mistranslation (30.7%), terminology (18.9%), and grammar (12.5%). All these errors can cause significant trouble, especially within patent translation because the Japan patent office is very strict.
Mr. Aoyama then listed several tools and methods used by survey participants to minimize mistakes: proofreading, CAT tools, term management, peer review, QA tools, and “other” (such as AI or text-to-speech); the most used techniques were proofreading, then CAT tools, then term management. However, Mr. Aoyama noted that peer review is not common, possibly due to time constraints or confidentiality issues, though he himself also uses AI to catch mistakes.
One of the most interesting parts of the presentation was details about the kinds of mistakes that are common in English-to-Japanese patent translation, such as romaji-to-kanji mistakes and particle mistakes, and terminology mistakes. One example used was the word “henzaiteki.” This spelling in romaji can result in two different words: 偏在的, meaning “uneven distribution,” and 遍在的, meaning “ubiquitous.” When typing using romaji-to-kanji, mistakes with homophones can result in serious translation errors.
Another mistake often seen was with particles, like は, が, に, etc. Even native speakers can make particle mistakes (which, as a non-native Japanese speaker, was comforting to hear). Therefore, in order to avoid particle translation mistakes, it is vital to know the whole sentence for full context. When to use は or が, or translating “not ~ but ~” were some examples given of common particle mistakes.
Of the other types of common mistakes, omission errors and inconsistency errors can be solved with CAT tools, terminology management, or AI; style and formatting issues can be solved by finding style guides either online or from a direct client. But terminology errors can require more care. Katakana words can cause issues, such as with レジスター, which could be either “resistor” or “register.” Similarly, English words may change meaning depending on the context; the word “solid” might be 忠実 or 個体; “residue” could be 残基, 残留物, 残渣, or 剰余. Or even a misunderstanding of a common phrase, such as “room temperature,” can be ambiguous, and therefore difficult to translate. In all examples, context becomes key to accurate translation, reinforcing that a translator must know the entire phrase and context of the sentence they are translating.
Mr. Aoyama ended his presentation with advice for fixing translation mistakes: Proofread multiple times, especially after a night has passed; consult with other, more knowledgeable professionals; review work and run spellcheck; read both source and target documents out loud; and give yourself as much time as possible between translating and proofreading. By using these techniques, one can minimize mistakes and provide the best possible translation.
Mr. Aoyama’s presentation illuminated the kinds of translation errors that any professional translator will inevitably run into on their own. The collected data and concrete examples of the various errors that can occur were extremely helpful, and the survey showed that every translator, no matter how experienced or skilled, makes mistakes. The examples of errors and the advice for avoiding them gave me a framework for how to think about the mistakes I may make in my own work. I hope to use this framework moving forward to create better translations.
Editors: Syra Morii, Tomoko Herbert
Leave a Reply