The Three Classes of Translation Errors and How to Reduce Them

By | March 23, 2020

For everyone out there doing translations––whether you’re an experienced career translator or a student just getting your feet wet––producing translations with a minimum of mistakes is surely high on your list of priorities.  

Notice I didn’t use the word “correct”. That’s because nearly all translations have a strong element of subjectivity and style in them. However, there are nevertheless indisputable errors that can arise at different stages of the process, and in this article I’d like to talk about those. While I personally work with the Japanese/English language pair, most of this article should apply to translation between any two languages.

But before I get into details, I want you to keep in mind that these classes of errors are relatively separate, and can be dealt with separately. Furthermore, to a certain extent the skills utilized at each stage can be improved separately.

My three categories of translation errors are based on the stage each occurs in: understanding the source text, translating, and editing/proofreading. I’ll now go over each of these.

Understanding the source text

Perhaps it’s obvious, but it doesn’t hurt to emphasize that superb translations come from a superb grasp of the source text––a term that refers to the text you are translating (which I’ll abbreviate as “ST”).

Understanding the source text means that you know the meaning that the author intended to convey to his or her readers, or at least have a confident guess at it. This understanding generally ideally does not involve translating everything to the source language in your head. I would even argue that you won’t be using just the target language either––I feel that true understanding requires abstract thought processes that cannot easily be expressed in words. I’m sure you’ve had an experience where you say to yourself, “Oh, now I get it!” but can’t necessarily explain it to another person, at least without significant effort to “put things into words”.

There are varying levels of scope when talking about understanding a text:

  • Word level: the meaning and nuance of individual words
  • Grammar level: the meaning of various grammar constructs
  • High level: what the author is really getting at, including explicit and implicit meanings

Fortunately, increasing your ability to understand texts in a foreign language over the long term is somewhat straightforward, if not time-consuming. It involves things like studying key areas including grammar and vocabulary by leveraging resources such as textbooks, classes, personal tutors, and online learning (blogs, etc.). But above all, you need to read (and do your best to comprehend) as many texts as possible, preferably with a good portion in the genre(s) that you plan to translate. Talking to others about a text you have read also can help you gain a deeper understanding of not just the text itself, but the words and phrases used.

But the great thing is that none of the above really applies specifically to translation, so your knowledge can be applied to other things, like conversation, writing, etc.

So how do mistakes creep in, assuming you are doing your best to research every part of the text you are uncertain about? 

There are a few sources of mistakes, but perhaps the most common one (besides careless mistakes) is when there is ambiguity about a word, phrase, or sentence, and simply reading grammar rules and dictionary entries doesn’t tell you the proper interpretation. The more years of experience you have with the source language, and the more years you have actively used it, the more likely your intuition has developed to the point where these types of mistakes are reduced to a small number. But for many of us who are not native speakers of the source language, I feel there can be areas of the ST that we lack confidence about, even extended research.

You can re-read the text several more times, and maybe even try a draft translation of it to see if we have a revelation about the proper meaning. But usually the only surefire way to improve your confidence in the interpretation of a passage is to talk to a native speaker or someone with significantly more experience than you in the target language.

Again, the great part about this is you don’t have to ask for help in terms of translation. In fact I would recommend staying in the source language when asking questions, and phrasing your questions in the form of “I think this means X, but perhaps it means Y? What do you think?” Don’t be too open-ended, and don’t say “Can you translate this for me?” Odds are that unless they have translation experience and are reasonably fluent in both languages, having the other person try to translate things will only confuse you and add uncertainty. When verifying your understanding, phrasing things in your own words (again in the source language) is important since you can choose words you have a good grasp of. If you ask someone to simply explain the meaning, they may end up with an explanation significantly more complex that the original text (I’ve actually had this experience several times before on Oshiete Goo).

There is another challenge here, in terms of understanding peoples’ strengths and weaknesses. Some people may have a good intuitive understanding of language but can struggle to explain to others in simpler terms. Others may have a lack of confidence in their language ability (even native speakers) and feel uncomfortable explaining things. Some people may be very strong in areas they are familiar with (say, science fiction) and of less reliable in other areas.

The more people you work with doing meaning validation, the more you’ll learn how to determine their strengths and weaknesses, and how to interact efficiently with them. 

For particularly difficult areas you can even ask several people for their opinions. But then you run into another problem: figuring how to resolve conflicts, since different people will likely have different interpretations of the text. Rather than try to synthesize all the info, I find it best to pick the feedback from a specific person and use that for the rest of the translation.

To briefly describe my process, when I go through an initial translation I mark areas for which I have doubts about in red, and then after another iteration or two I pair those down to the ones that I’ve been unable to figure out by researching on my own. If there is a large number of them, I try to prioritize those that really impact the story or I’m clueless about. While it would be great to ask someone to check your understanding of every sentence, it’s just not feasible in most cases (the exceptions being if they are a close friend, relative, or perhaps if you are paying them).

If you are translating text from an author who is still alive, you may have an opportunity to ask them questions directly. I’ve done this on occasion, and have gained some important insights from it. However, I would caution that you only do this for very critical areas, and you be extremely careful with your phrasing. Regardless of your experience level, authors might assume you are a professional and may not appreciate being asked questions about Japanese grammar or vocabulary.

You may be able to uncover things that are not explicitly (or even implicitly) in the source text, especially if you are able to speak directly with the author. This is important knowledge, but during translation it should be used wisely after careful consideration. For example, in most cases I would not suggest adding extra details to the text based on feedback from the author or your own interpretation, unless they are cultural or linguistically related and readers of the source text might have inferred such information on their own. Even if you feel giving extra context is warranted, it’s generally safer to use translation notes instead of modifying the text itself. 

On a final note, while it’s possible that by lucky chance errors at this stage are corrected in a later stage, there’s a good chance they will end up in the final product. And your readers (including proofreaders) are not likely to catch them.


Of course, translation of a text from one language to another is the high-level goal of our efforts, but here by “translation” I mean the process of taking sentence-by-sentence of a (presumably well-understood) source text and attempting to render it into your target language. The key point is that what you write is not necessarily going to be the final product since there will be a stage of proofreading and editing after this. This translation stage is by far the trickiest, because it involves the most subjectivity of all the stages, and is the stage that is the most difficult to involve others in.

Assuming you have 100% comprehension of the ST, what kind of mistakes can be made in translation? 

There are two subcategories of these types of errors. The first is errors that come about because a word/phrase/grammatical construct in the source language doesn’t exist in the target text, and the translator must try to find something close enough. While things aren’t black and white––there isn’t one “correct” translation––for any given phrase surely some translations are closer to the ST than others. The frequency of situations where there is no obvious equivalent word/phrase depends on how related two languages are, but for Japanese and English this is quite common.

Using tools like dictionaries and thesauruses can help here, but I feel the most important thing is that the translator has a good grasp of the target language. Reading and writing––and, to a lesser extent, speaking and listening––as much as you can in that language are really the only ways to truly develop the needed skills. 

The second subcategory of errors in the translation stage is simply errors where you omit a portion of the ST. The longer and the more complex the sentence structure is, the more likely this is to occur. Fortunately, this error can be drastically reduced by making sure when you are iterating through your translation you are not only checking the quality of the target text (that is the next stage), but also double-checking against the ST. Frankly, this can be a tedious process, but if you can line things up such you can see both paragraphs (one in each language) on your screen at the same time, it will make things much easier.

This stage is also the hardest to have someone check your work because it would require someone being bilingual (and ideally having translation experience). While you can probably find people to help out, the majority of the time you are probably on your own here.

Proofreading / Editing

This last stage involves going through the translation draft and looking for typos (generally objective errors), as well as adjusting wording and sentence structure to make the sentence read naturally (partially objective, partially subjective).

While it may seem there are some things in common with the previous step, the key part about the proofreading/editing step is that it can occur largely without reference to the source text. 

Once you (the translator) has got to this stage, you have probably gone through the source text several times already, and maybe even started memorizing portions of it. So even if you read the translation draft in isolation, you will have a strong bias (unless you wait until you forget things which may take months or longer). On the other hand, readers of your translation will generally have no information about the ST and will be reading it with a fresh mind.

What this means is that the translator is not going to be the best person to proofread their translation draft, and in fact depending on the publisher some translation projects will have dedicated editors to help with this process. But even if you are doing smaller projects or self-publishing, it would behoove you to find one or more people to help proofread, whether for pay or for gratis.

There are two types of people that can help you: everyday readers (so-called “beta readers”), and people who have formal experience in proofreading and editing. Of course it’s great if you can get a professional editor to help check your translation, but even a handful of beta readers can make a big difference in finding things like typos and awkward phrasing. Keep in mind that just like with the ST meaning verification, you should have the final judgment on what is best for the translation. (On second thought, the final judgment could be given to the publisher/editor in some circumstances.)

Now, just because it’s essential to have another pair of eyes looking at your translation doesn’t mean you shouldn’t be doing your best to proofread your own work. You can also use tools like Grammarly to help find some of the more common errors, but there are limitations to what such tools can do for you.

For my translation process, it’s a little less clear cut than I am describing here; I don’t finish the translation stage and then go to the editing/proofreading stage. Rather, I go through a series of iterations where I gradually pay less attention to the source text. By the end, I have removed the ST from my document and am only staring at the target text, which is a translation with some degree of polish. While portions of the ST will probably spring to mind as I read, I try to focus on the target text as if I was an unbiased reader. 

Because the proofreading/editing stage is mostly separate from the source language, any knowledge about editing in the target language will be helpful. So you can take courses on editing (or writing in general) and that will improve the quality of your translations.


In reality, there are some connections between these steps. For example, someone who is strong in both translation and writing/editing may be able to essentially do both those stages at the same time, though the majority of cases will require more than one iteration for a good quality level.

Also, some translators may wish to reflect certain aspects of the ST author’s style in the target text. This can include things like sentence length, word nuance, or strong imagery. Some of these things may carry over as part of the normal translation process, but others may require an end-to-end look at both texts, and conscious decisions involving what aspects of the ST author’s style to try and incorporate. Often the way this works in practice is the translator makes certain stylistic decisions as they go (for example, maintaining paragraph boundaries), and then tries to follow the precedents they have set consistently. 

Finally, editing changes during the proofreading stage ultimately need to be evaluated in light of the ST, at least in some cases. If you imagine someone translates a short story and then passes the translation off to an editor who knows nothing about the ST, he or she may make changes that sound great in the target language but distinctly deviate from the ST’s meaning. I’ve heard this does happen in practice, and ultimately it’s up to the publisher as to what they decide is best. I get the feeling that more often than not publishers would prefer a natural but less-literal rendering to one that is accurate but unnatural. (That in itself is a major philosophical debate in the world of translation…)

I hope this discussion about my three categories of translation errors has been beneficial to you. Hopefully we can all gradually increase the quality of our translations while reducing the time it takes to produce them. And don’t forget to have fun! (:

(By the way, if you are interested in learning more about translation, you can see my Youtube series TransLiterary Lab here where I translate short excerpts of literary works.)

[Note: Featured image of “1/2/3” taken from]

(Visited 409 times, 1 visits today)

Leave a Reply

Your email address will not be published.