A client asked me to proofread machine translation

A few weeks ago, a client asked me if I could review a text that had been translated by a machine.

I said no, but I also said that I could translate it from scratch.

Why? Because machine translation comes with too many issues:

🚫 Structural rigidity : it follows the source structure too closely, which makes the target text sound less natural.

🚫 No nuance : expressions, metaphors, and cultural references are often mistranslated, the meaning of the text is then lost.

🚫 Gender mix-ups : it randomly switches between genders when translating from a non-gendered language to a gendered language.

🚫 Error blindness : it doesn’t spot mistakes in the original text, so it can’t help you improve your text.

🚫 No terminology consistency : it doesn’t use your glossaries, so it doesn’t know how you normally talk about specific concepts.

🚫 No respect for tone or style : it can’t follow guidelines on tone.

🚫 Inaccurate link handling : it doesn’t adapt links when needed, which means the texts cannot always be fully used by the target audience.

🚫 Literal translations of quotations and names : even when official translations exist, it results in a lack of homogeneity in your communication.

🚫 Concept inconsistency : it translates the same idea with different words throughout a text, so your reader won’t know if you are talking about the same thing or if you are introducing a new concept.

🚫 Random additions : sometimes machine translations “creates” meaning when it doesn’t know how to translate something, which means it randomly adds information that doesn’t exist in the source text.

And when it comes to inclusive language, it’s even worse:

❌ It doesn’t recognize problematic expressions so you take the risk of having a problematic target text.

❌ It can’t apply inclusive language methodologies, so your texts are going to follow sexist grammar rules.

❌ It can’t provide inclusion-related advice, so if you missed some problematic language in English, it can’t let you know.

That’s why instead of fixing a flawed text, I prefer to deliver a high-quality translation from the start—one that’s accurate, inclusive, and aligned with my clients’ needs.

❓ Have you ever had to deal with machine translation fails?