Professional post-editor for machine translation and ISO 18587 post-editing

A professional post-editor reviews, corrects and improves the machine-translation output by comparing it with the source text. Their work is not a spell check or a quick reading of an automatic text. In a human post-editing project, the post-editor must check meaning, terminology, style, omissions, additions, numbers, format and suitability for the intended audience.

Professional post-editor working on bilingual texts and machine translation output

The role of the post-editor is central to ISO 18587. The standard focuses on full post-editing by a qualified linguist of machine-translation output and defines the competences needed for this task. Without a qualified post-editor, machine-translation output may remain fluent but unreliable.

LinguaVox works with professional post-editors for projects in which machine translation with post-editing is viable. Before assigning the work, we assess the text, language pair, field, terminology and intended use.

What a post-editor does

A post-editor compares the machine-translation output with the source text and decides what must be corrected. This includes meaning errors, omissions, additions, terminology problems, grammar, syntax, style, punctuation, formatting and consistency.

The post-editor also decides when an automatic segment can be corrected and when it must be rewritten or translated again. This judgement is one of the main differences between professional post-editing and superficial correction.

Why fluency is not enough

Modern machine translation can produce fluent sentences. That fluency can be misleading. A sentence may sound natural and still omit a condition, invert a relation, use a wrong term, alter a number or introduce information that does not appear in the source text.

For this reason, a post-editor must be cautious with plausible output. The most dangerous errors are not always the most visible. They are the errors that look correct at first sight but change the message.

This is especially important in technical, legal, medical, financial, software and corporate content. A mistranslated warning, a changed obligation, an inconsistent product term or a damaged variable can create practical problems.

Competences required under ISO 18587

ISO 18587 describes the competences expected from post-editors. They need translation competence, linguistic and textual competence in the source and target languages, research competence, cultural competence, technical competence and subject-matter awareness.

Translation competence is essential because the post-editor must understand the source text and produce an accurate target text. Linguistic competence is needed to write naturally in the target language. Research competence helps resolve terminology and subject-matter doubts.

Technical competence is also important. Post-editors often work in CAT tools, with translation memories, terminology databases, machine-translation output, quality checks, tags and file formats that must not be damaged.

Typical errors detected by a post-editor

A professional post-editor looks for more than grammar mistakes. They check whether the automatic output has omitted information, added content, mistranslated terms, changed logical relations, damaged formatting or produced inconsistent style.

Common errors include false friends, literal translations, wrong terminology, incorrect numbers, altered units, mistranslated conditions, missing negatives, inconsistent product names, broken tags and unnatural sentence structure.

Some errors are language-specific. Others depend on the field. A software interface, a technical manual, a medical leaflet and a product catalogue do not generate the same risks. This is why subject-matter knowledge matters.

Post-editor and translator: related but not identical roles

A translator produces the target text from the source text. A post-editor starts from machine-translation output and evaluates whether it can be corrected. Both roles require translation competence, but the cognitive task is different.

The post-editor must resist two opposite risks. The first is accepting too much machine translation because it sounds fluent. The second is rewriting everything unnecessarily because of personal style preference. Professional post-editing requires balance.

In full post-editing, the final result should be comparable to human translation. That does not mean changing every sentence. It means making all changes needed for the text to be accurate, natural, consistent and fit for purpose.

Post-editor and reviewer: what is the difference?

A reviewer normally checks a human translation. A post-editor checks machine-translation output. The distinction matters because machine translation produces different types of errors from human translation.

Human translations may contain misunderstandings or style issues, but machine translation can create fluent hallucinated relations, terminology shifts, omissions, additions or false consistency. These errors require specific attention.

A reviewer may also be involved after post-editing in some projects, depending on the agreed quality workflow. However, the post-editor remains responsible for the bilingual correction of the automatic output.

How LinguaVox assigns post-editors

LinguaVox assigns post-editors according to language pair, subject matter, file type, expected quality and project risk. A post-editor for software strings does not necessarily fit a legal contract, and a technical manual requires different attention from a marketing page.

The project manager also checks whether the post-editor has the necessary resources: source files, machine-translation output, glossary, translation memory, style guide, previous translations and client instructions.

This preparation reduces uncertainty and improves consistency. It also helps the post-editor focus on real errors rather than guessing the client’s terminology or intended style.

Working with post-editors in multilingual projects

In multilingual projects, post-editors need common instructions. Product names, warnings, interface labels, terminology and formatting rules should remain consistent across languages unless there is a justified local adaptation.

A central glossary and shared project brief help avoid fragmentation. Feedback from one language may also reveal problems that affect other languages. If several post-editors report the same issue, the source text or machine translation settings may need adjustment.

LinguaVox coordinates these elements so that post-editing is not treated as isolated correction by language. The aim is a coherent multilingual result.

Linguistic decisions on meaning, terminology, style and text consistency

When a post-editor should reject a segment

A post-editor does not have to preserve the machine-translation output. If a segment is misleading, incomplete, ungrammatical, terminologically wrong or stylistically unusable, it should be rewritten or translated again.

This is particularly important when the automatic text creates a false sense of quality. Keeping a fluent but wrong sentence is more dangerous than rewriting it. The post-editor must make decisions based on accuracy, not on the appearance of the output.

In some projects, repeated rejection of segments indicates that the workflow itself may be unsuitable. If too much of the output must be rewritten, human translation may be more efficient.

Tools used by post-editors

Post-editors may work with CAT tools, translation memories, terminology databases, machine translation engines, quality assurance checks and client-specific instructions. These tools help manage consistency and file structure, but they do not replace professional judgement.

Quality assurance tools can detect numbers, tags, repeated segments, terminology and formatting issues. They cannot fully determine whether the meaning is correct or whether the style fits the intended audience.

The post-editor remains responsible for the linguistic and conceptual decisions. Technology supports the task; it does not perform the professional evaluation by itself. You can see more context in the section on machine translation tools.

What a post-editor needs before starting

A post-editor needs more than a source text and a machine-translation output. They need to know the purpose of the document, the target audience, the required level of post-editing, the terminology rules and any formatting constraints. Without that context, the post-editor must make decisions that may not match the client’s expectations.

Useful resources include glossaries, previous translations, translation memories, style guides, product names, screenshots, reference websites and client instructions. Even a short explanation of how the text will be used can prevent wrong assumptions.

This preparation is especially important in ISO 18587 projects because the task is not only to make the target text readable. The final document must be accurate, consistent and suitable for the agreed use.

Subject-matter knowledge

Post-editing requires subject-matter awareness. A post-editor does not need to be an engineer, lawyer or doctor in every project, but they must understand the type of content and know when terminology or meaning requires research.

In technical documentation, a common word may have a precise function. In legal content, a modal verb or condition can change the effect of a sentence. In software, a label may need to match an interface element. In medical or pharmaceutical content, terminology and risk level require particular caution.

For this reason, LinguaVox does not assign post-editing only by language. The subject matter, document type and risk level also influence the choice of professional.

Managing style without over-editing

One of the difficult parts of post-editing is avoiding unnecessary rewriting. A post-editor must correct what is needed, but should not change every sentence just because another formulation is possible. Over-editing can increase cost and reduce consistency without improving the document.

At the same time, under-editing is risky. A fluent sentence may be wrong, unnatural for the target audience or inconsistent with the client’s terminology. The post-editor must distinguish between a harmless stylistic variation and a real quality problem.

Post-editor feedback and continuous improvement

Post-editors can provide useful feedback beyond the corrected file. If the same problem appears repeatedly, it may reveal an issue with the machine translation engine, the source text, the glossary or the project instructions.

For example, the post-editor may notice that a product component is repeatedly mistranslated, that variables are being damaged, that the source text contains ambiguous sentences or that the same term is translated in several ways. This information can improve future batches.

In recurring projects, LinguaVox can use this feedback to update glossaries, clarify instructions or recommend pre-editing of the source text. This makes post-editing more efficient over time and reduces repeated corrections.

Post-editing in multilingual teams

In multilingual projects, several post-editors may work on different target languages. They need shared instructions so that terminology, product names, warnings, UI labels and formatting rules remain consistent.

A problem detected in one language can be relevant for others. If a source sentence is ambiguous, every post-editor may face the same risk. If a term is not in the glossary, each language may solve it differently. Coordination helps avoid that fragmentation.

The project manager plays an important role here. They collect doubts, update instructions and ensure that language-specific decisions do not damage the consistency of the overall project.

Frequently asked questions about post-editors

What is a post-editor?

A post-editor is a language professional who corrects machine-translation output by comparing it with the source text.

Is a post-editor the same as a translator?

The roles are related but not identical. A translator produces the translation from the source. A post-editor corrects and evaluates machine-translation output.

Can any bilingual person post-edit machine translation?

No. Professional post-editing requires translation competence, language competence, research skills, technical competence and knowledge of machine translation errors.

What does a post-editor check?

They check meaning, terminology, omissions, additions, grammar, style, numbers, units, tags, formatting and consistency.

When should machine translation be rejected?

It should be rejected when the output is misleading, incomplete, too literal, terminologically wrong or inefficient to correct.

Does ISO 18587 define post-editor competences?

Yes. ISO 18587 includes requirements related to the competences and qualifications of post-editors.

Request a professional post-editing service

Send the source file, machine-translation output if available, language pair, intended use and terminology resources. LinguaVox will assess whether a professional post-editor can work efficiently on the material.