Mistakes in Word-by-Word Translation and Degrees of “Rightness” in Translation
by Chad Richardson
Many people think that translation is a simple, mechanical process. You take each word in the source language, figure out its meaning in the destination language and then write down that word in your translation. But if this was how it worked, then anyone would be able to translate from one language to another with the help of a dictionary.
Different Languages, Different Structures
If you have ever tried word-for-word translation, you’ll know that the result tends to be quite ludicrous. This is because different languages function in different ways. They have different structures. For example, to negate a sentence in English, you only have to use the word “not.” So “He can speak well” becomes “He can not speak well.” But to negate the same sentence in French, you have to add “ne” before the verb and “pas” after the verb. So “Il parle bien” becomes “Il ne parle pas bien.” Hence, the structure of the language is different.
Examples of Word-by-Word Translation
Sometimes, you might get lucky in word-for-word translation. For example, “He can speak well” becomes “Il peut parler bien” which does make sense.
But, in other cases which are more complex, the result of word-by-word translation makes no sense at all. For example, consider the first sentence of this blog post: “Many people think that translation is a simple, mechanical process.” Using word-for-word translation, one comes up with “Beaucoup gens pense cette traduction est une simple, mecanique processus.” This word-by-word translation was just done using Google translate and it is quite obvious that it was done by someone who doesn’t speak French.
Mistakes in Word-by-Word Translation
What are the mistakes in the above sentence? To begin with, the word “beaucoup” which means “many” should always have “de” following it. Plus, the verb “pense” does not agree with the subject; it should be “pensent” which is plural. “Cette” is an incorrect translation of “that”; it should be “que.” And there are many other grammatical oddities in this sentence.
Degrees of “Rightness” in Translation
The following sentence would be a far better translation: “Beaucoup de gens pensent que la traduction est un processus simple et mécanique.” Even this sentence, though grammatically sound, may come across as awkward and it might be better to say, “Beaucoup de gens pensent que la traduction est simple et mécanique.”
In either case, it’s important to remember that nothing is 100% right or wrong when it comes to language. There are degrees of rightness and our aim is to get your translation as right as possible so that when someone who is fluent in the destination language reads it, it makes intuitive sense to them.
Contact us for the best possible translation of your document.
Related Articles
What are spell check exceptions in localization?
Localization spell check exceptions are whitelisted parameters within a translation management system that prevent approved brand names or proprietary terms from being flagged as errors. By configuring these exceptions, companies ensure their language operations remain scalable and efficient while protecting the unique terminology that defines their global brand identity across 200+ supported languages. Why do…
How Does a GitHub Connector Work in Localization?
A GitHub Connector is the technical bridge that integrates your code repository directly into a continuous translation process. By automating the exchange of resource files – such as JSON, XML, XLIFF, or STRINGS – it eliminates manual file handling, reduces human error, and ensures that translation happens in parallel with the development sprint rather than…
Can AI Detect AI? Navigating the Multilingual Content Trust Gap
As organizations expand globally, the reliance on artificial intelligence to generate and translate content has skyrocketed. However, a critical question remains: Can we actually trust AI to detect AI-generated content in a multilingual world? Recent research from Pennsylvania State University, MIT Lincoln Laboratory, and Trinity College Dublin involving the BLUFF benchmark – a dataset covering…
here
for you
We’d love to learn more about your translation and localization needs.