Can a translator with a C2 certificate still produce English that reads like it was written by a machine? Yes, and it happens every day. That gap between “very good English” and “English that works as a translation” is exactly what English for translators is about. It is not a general language course with harder vocabulary. It is a set of specific skills: knowing which collocations survive in print, spotting a false friend before it reaches the client, and understanding why a sentence that is grammatically perfect can still sound wrong to a native reader.

This article looks at what those skills are, how training programs build them, where the standard advice gets shaky, and what actually works in practice.

What “English for Translators” Actually Covers

The label gets used loosely. In practice, it covers four overlapping areas.

Translation into English. Most of the work happens here, with translators rendering from a source language into English as a second or third language. Professional bodies have long advised translating only into your mother tongue, and that convention still shapes how training is structured. It is guidance rather than law, and it is increasingly qualified for certain language pairs and specialist domains.

Register and genre control. A legal contract, a patient leaflet and a fashion blog need three different Englishes. Translators have to switch between them without slipping.

Text-level editing. Revision and review are treated as separate tasks from translation, with their own competence requirements, and English-for-translators training reflects that split.

Tool and workflow literacy. Modern courses fold in machine translation post-editing, terminology management and corpus use, because those are now part of the job rather than optional extras.

Why Good General English Is Not Enough

A translator can read a novel in English without effort and still freeze on a product label. The reason is that translation puts pressure on parts of the language that everyday fluency never tests.

General English rewards comprehension. Translation rewards production under constraint: the meaning is fixed by the source text, so the translator cannot simply rephrase until something sounds nice. Every word choice has to carry the original meaning and land naturally in English at the same time.

The Common European Framework of Reference, the scale behind labels like B1 and C2, was designed for general language learners, not translators. There is no official CEFR level for translation. When a job ad says “C1/C2 English required,” that is convention, not a standard, and it tells you almost nothing about whether the candidate can write a clean press release.

The Competence Framework Behind Good Training

The international standard for translation services, ISO 17100:2015, sets out what a competent translator is expected to bring. Alongside linguistic and textual competence in both the source and target language, it names cultural competence and the technical know-how to handle the tools of the trade. It also separates translation, revision and review as distinct tasks with distinct requirements.

That structure matters for anyone choosing a course. A program that only drills grammar is training a language learner. A program built around the standard is training someone who can translate, then revise someone else’s work, then review a finished text against a brief.

A practical checklist for judging a course:

• Does it teach revision as a separate skill, with practice texts and a marking scheme?

• Does it cover register and genre as well as vocabulary?

• Does it include corpus work, so you learn to check a collocation instead of guessing?

• Does it address machine translation post-editing as its own discipline?

• Does it discuss language direction honestly, including the limits of translating into a second language?

False Friends, Collocations and the Translationese Trap

False friends are the famous problem: words that look alike across two languages but mean different things. They are easy to teach and easy to test, which is why they dominate classroom time.

Collocations are the quieter problem. English has strong preferences about which words keep company with which. A translator can pick a synonym that is perfectly correct in meaning and still produce a phrase no English speaker would write. Dictionaries list meanings; they are much weaker at showing which combinations are normal.

Translationese is the umbrella term for features that mark a text as translated rather than originally written. Corpus-based translation studies has treated it as a research area since at least the 1980s and 1990s, and the symptoms are recognizable: over-explanation, repeated sentence structures, unusual word order carried over from the source, and a fondness for Latinate vocabulary where plain verbs would do.

The fix is not more grammar. It is exposure to large amounts of real English and the habit of checking. Corpus tools such as the British National Corpus, completed in 1994, and the Corpus of Contemporary American English, launched in 2008, let a translator type in a suspected phrase and see whether it actually occurs, and in what register. That single habit, checking rather than trusting your ear, prevents more errors than any rulebook.

Reading Level and Register: Matching the Brief

Two translations can both be accurate and still differ in quality, because accuracy is only half the job. The other half is fit: does the English match what the client asked for?

Text type	Typical register	Common trap
Legal contract	Formal, repetitive by design	“Improving” deliberate repetition
Patient information leaflet	Plain, short sentences	Over-formal vocabulary
Marketing copy	Idiomatic, punchy	Literal renderings of slogans
Academic abstract	Nominal, hedged	Copying source syntax
Technical manual	Imperative, consistent terms	Synonym drift across sections

The last row is where careful translators earn their fee. In a manual, the same part must carry the same name on page 3 and page 90. Swapping in an elegant synonym for variety is a defect, not a stylistic improvement.

Plain-language guidance is useful here. Clear-writing resources aimed at public-sector English, including the European Commission’s long-running “How to write clearly” booklet, are standard reference material for translators working into English, and they are a good antidote to the nominal style that creeps in from many source languages.

Translating Into English: The Honest Conversation

The most contested question in the field is whether a translator should work into a language that is not their first. The traditional answer, endorsed for decades by professional bodies, is no. The reasoning is sound: a non-native writer can miss register, idiom and cultural nuance in ways that are hard to self-detect.

The practical answer is more complicated. Certain language pairs, certain domains and certain markets rely heavily on translators working into English, and quality varies enormously within that group. What separates the strong from the weak is rarely raw vocabulary. It is whether the translator knows their limits, uses a native-speaker reviser when the stakes are high, and checks rather than assumes.

Institutional demand shaped this reality. English-language translation volume in the EU institutions grew sharply after the 2004 and 2007 enlargements, when English became the most common relay and second language among incoming staff. That created a large, permanent market for high-quality translation into English by non-native speakers, and it raised the bar for what “good enough” means.

Where Machine Translation Fits

Post-editing machine translation output is now a standard competence area, with its own international standard, ISO 18587:2017. Ignoring it in a training program is like teaching typing on a typewriter.

The skill is not “fix the robot’s grammar.” It is judging when the raw output is usable, when it needs light editing, and when it needs to be thrown out and retranslated from scratch. Post-editors also need to resist a specific temptation: accepting a fluent-sounding sentence that quietly drops a qualifier from the source.

A workable rule of thumb from practice: read the machine output once without editing, decide whether the sentence structure is salvageable, and only then start changing words. Editors who fix word by word tend to preserve bad structure and miss meaning errors.

Practical Habits That Separate Strong Translators

Experience in the field points to a handful of habits that matter more than talent.

Build a personal phrase bank. Every time you resolve a tricky collocation, save the sentence. Over a year, this becomes a reference no dictionary can match.

Read your target text aloud. Awkward rhythm is the fastest detector of translationese. If you run out of breath, the sentence is too long.

Separate drafting from revising. Translate first, then edit in a different session. Doing both at once produces text that is half-checked and half-guessed.

Check the boring words. Prepositions, articles and verb patterns cause more errors in translated English than rare vocabulary ever does.

Keep a list of your own recurring mistakes. Everyone has three or four. Knowing yours is worth more than a shelf of style guides.

Ask what the text is for. A translation read once on a website and a translation read in a courtroom need different levels of polish, and clients rarely say which they need unless asked.

If your source language pairs with English in ways that create specific structural habits, it helps to study how other translators handle the same problem. Resources on hair types and texture, for instance, show how much descriptive precision depends on shared vocabulary rather than literal equivalence, whether the topic is 4c hair or the difference between 2b hair and looser waves. Terminology work in any specialist field follows the same logic: agree on terms, then stay consistent, as translators do when handling a series like The row.

How to Build Your English Between Assignments

Formal courses help, but most improvement happens in the gaps. A realistic weekly routine:

• Translate one short text from your strongest source language, then leave it for 24 hours before revising.

• Revise someone else’s translation, or have yours revised. Revision is where the real learning happens.

• Check five doubtful collocations in a corpus and note what you find.

• Read 2,000 words of high-quality original English in a register you do not usually handle.

• Post-edit a machine translation of a text you have already translated yourself, and compare the two.

That last exercise is uncomfortable and extremely effective. It shows exactly where the machine beats you and where it quietly gets things wrong.

Conclusion

English for translators is a craft skill, not a certificate. It rests on knowing the difference between correct and natural, on checking your instincts against real usage, and on being honest about language direction and the limits of your own ear. The standards exist, from ISO 17100 on translator competence to ISO 18587 on post-editing, and they are worth reading rather than guessing at. But the daily work is smaller and more repetitive: one collocation at a time, one revision at a time, one saved sentence at a time. Translators who treat English as a living system to be maintained, rather than a box already ticked, are the ones whose work reads as if it were written, not converted.

Frequently Asked Questions

Do I need to be a native English speaker to translate into English?

No, but you need to be honest about your limits and build checks into your workflow. Professional bodies have long advised translating only into your mother tongue, and that guidance exists for good reasons. In practice, many translators work into English as a second language in specific domains, and quality depends far more on revision habits, corpus checking and knowing when to bring in a native reviser than on birthplace.

Is a C1 or C2 English certificate enough to work as a translator?

Not on its own. The CEFR scale was designed for general language learners, and there is no official CEFR level for translation. A certificate shows you can understand and use English broadly. It says nothing about whether you can control register, avoid translationese or revise another translator’s work, which are the skills clients actually pay for.

What is translationese and why does it matter?

Translationese describes features that make a text read as translated rather than originally written: unusual word order carried over from the source, over-explanation, repeated sentence patterns, and Latinate vocabulary where plain verbs would be more natural. It matters because readers notice it even when they cannot name it, and it undermines trust in the text. Corpus checking and reading widely in the target register are the most reliable fixes.

How do corpora help translators improve their English?

Corpora let you test a phrase instead of guessing. Large collections such as the British National Corpus and the Corpus of Contemporary American English show whether a word combination actually occurs, how often, and in which register. That turns a vague feeling about what sounds right into evidence, which is especially useful for collocations that dictionaries handle poorly.

What is machine translation post-editing, and is it a separate skill?

Yes, and it has its own international standard, ISO 18587:2017. Post-editing means working on machine output rather than translating from scratch, and it requires judgement about when the output is usable, when it needs light editing and when it should be discarded. The hardest part is catching fluent sentences that quietly drop meaning from the source.

How much does language direction really matter in practice?

It matters, but it is not the only factor. Translating into a non-native language is widely discouraged as a default, yet large markets depend on it, particularly where English serves as a relay language. The practical question is whether the translator has a reliable revision process, domain expertise and the discipline to flag uncertainty. Those factors predict quality better than native-speaker status alone.