Editing a Translation Without the Original
I remember sitting in a windowless office in Bloomsbury ten years ago, staring at a manuscript that had been “professionally translated” from German, only to find it read like a manual for a mid-range dishwasher. The client had paid a premium for the translation, yet the prose was so stiff it practically creaked. People often think that once the translation is done, you just need a quick proofread to catch a few typos, but that is a dangerous misconception. If you don’t actually understand how to edit translated text as a distinct, structural discipline, you aren’t editing; you’re just rearranging the furniture in a house that’s built on a swamp.
I’m not here to give you a lecture on the beauty of linguistics or the soul of a language. I’m here to talk about the work. I will show you how to spot the “translationese” that kills reader engagement, how to manage your time when the syntax feels fundamentally broken, and how to charge a rate that actually reflects the extra cognitive heavy lifting this requires. We’re going to look at the mechanics of the job so you can stop wasting hours on drafts that were doomed from the first sentence.
Why Post Editing Machine Translation Is Bleeding Your Budget Dry
The math behind post-editing machine translation (PEMT) is often presented as a miracle of efficiency, but in practice, it’s frequently a mathematical illusion. Clients see a lower per-word rate for machine output and think they’ve found a loophole in the budget. What they don’t see is that they aren’t paying for a translation; they are paying for a forensic reconstruction. When you are tasked with fixing a raw MT output, you aren’t just smoothing edges—you are often fighting a losing battle to preserve maintaining source text meaning against a machine that has hallucinated a completely different concept.
I’ve seen projects where the “savings” from using AI were swallowed whole by the sheer volume of hours required for a proper translation quality assurance process. If the machine misses the subtle shift in tone or fails to grasp a specific idiom, a human editor has to untangle the knot from scratch. You end up spending more time deconstructing errors than you would have spent writing from a clean slate. In the end, you aren’t saving money; you’re just paying more for a much harder job.
The Linguistic Accuracy Myth and What Clients Actually Pay for
Clients often come to me with a slip of the tongue: “I just need someone to check the linguistic accuracy in translation.” They think they’re paying for a spellcheck, but if you’re only looking for typos, you’re missing the point of the entire engagement. Accuracy isn’t just about whether the verb agrees with the subject; it’s about whether the soul of the sentence survived the trip across the border. You can have a sentence that is grammatically perfect but feels like it was written by a very polite robot with no pulse.
When I price a project, I’m not just counting commas. I’m billing for the cognitive load of maintaining source text meaning while simultaneously hunting for those invisible errors that make a reader pause. A client isn’t actually paying for a translation quality assurance process; they are paying for the absence of friction. They want a text that reads as if it were born in the target language, not one that wears its foreign origins like a heavy, ill-fitting coat. If the reader notices the translation, you haven’t finished the job.
The Editor’s Checklist: Five Ways to Stop Polishing a Turd
- Hunt for the “translationese” rhythm. Even if the grammar is technically perfect, if the sentences all follow the same stilted, predictable cadence of the source language, the reader will feel it. You aren’t just fixing errors; you are re-injecting a natural human pulse into a text that currently feels like it was assembled by a very polite robot.
- Watch the idioms like a hawk. A machine might translate a metaphor literally, but a human knows that “it’s raining cats and dogs” doesn’t work in every culture. Your job is to decide whether to find a local equivalent or to flatten the metaphor entirely so it doesn’t trip the reader up.
- Check the terminology consistency against the clock. If you’re working on a technical manual or a niche non-fiction piece, the client isn’t paying you to be a poet; they are paying you to ensure “actuator” doesn’t become “lever” halfway through page forty. This is where the real hours go, and it’s where you earn your day rate.
- Scrutinize the tone, not just the meaning. A translation can be 100% accurate and still be a total failure if the source was a cheeky, irreverent blog post and the output reads like a legal deposition. If the “voice” is missing, you aren’t editing; you’re just proofreading, and you should charge accordingly.
- Don’t fall into the “correction trap” during the first pass. If you spend twenty minutes agonizing over a single semicolon in the first paragraph, you will never hit your word count target for the day. Get the structural flow and the major linguistic shifts sorted first; save the microscopic nitpicking for the final sweep once you actually know where the text is going.
The Bottom Line on Translation Editing
Stop treating post-editing like a quick skim; if you aren’t hunting for the structural awkwardness that makes a reader trip, you aren’t editing, you’re just proofreading, and you’re likely undercharging for the actual labor involved.
Understand that clients aren’t paying for “accuracy”—the machine can handle the literal meaning—they are paying for flow, tone, and the removal of that unmistakable “translated” scent that makes a text feel lifeless.
Protect your margins by distinguishing between a light MTPE (Machine Translation Post-Editing) task and a full stylistic rewrite; if the source text is a mess, don’t let a client bully you into a low per-word rate for work that requires a heavy-duty developmental edit.
The Ghost in the Machine
“If you’re hiring an editor to simply check if a machine got the nouns right, you aren’t paying for an editor; you’re paying for a glorified proofreader, and frankly, you’re paying too much for it. A real editor isn’t there to fix the grammar—they’re there to find the soul that the translation accidentally scrubbed out of the prose.”
Cressida Farrow-Bassey
The Bottom Line on Translation
At the end of the day, editing translated text isn’t about checking if the words exist in a dictionary; it’s about ensuring the soul of the original hasn’t been lost in a digital meat grinder. We’ve established that relying on raw machine output to save a few cents per word is a recipe for professional bankruptcy, both in terms of your reputation and your client’s bottom line. You cannot simply skim for grammatical errors and call it a day. If you aren’t hunting for those subtle, rhythmic inconsistencies that signal a non-native origin, you aren’t editing—you’re just proofreading a ghost. To do this job right, you have to treat the text as a living entity that requires structural integrity and cultural nuance, not just a linguistic puzzle to be solved.
If you find yourself feeling exhausted by the sheer volume of mediocre, machine-generated content flooding your inbox, remember that your value lies in the things a processor cannot replicate: judgment, irony, and empathy. The industry is shifting, yes, but it is also stripping away the fluff to reveal what actually matters. Don’t compete with the machines on speed; compete with them on substance and precision. When you charge a premium for your expertise, you aren’t just paying for your time; you are charging for the guarantee of human connection. That is a service that will always have a market, provided you have the discipline to demand it.
