Many companies reach the same point: a large volume of content to translate, a tight budget, and a deadline that does not allow for delays. The question that follows is practical: does machine translation post-editing solve the problem, or is human translation the only responsible option? The answer depends on the content type, the end use, and the risk attached to getting it wrong.
What MTPE is and how it differs from human translation
MTPE stands for Machine Translation Post-Editing. The process starts with a machine translation engine, such as DeepL, Google Translate, or a domain-specific neural system, which produces a draft. A certified human translator then reviews that draft, corrects errors, adjusts register, and ensures terminological consistency.
There are two recognised levels of post-editing:
- Light post-editing: the output is fit for internal use or informational purposes. Stylistic precision is not required; the content simply needs to be intelligible and accurate at the factual level.
- Full post-editing: the result must be equivalent to a translation produced from scratch by a human translator. This applies when the text will be published or distributed externally.
In pure human translation, the translator works directly from the source text, without a machine-generated draft. The process is slower and typically costs more per word, but it offers greater control over tone, structure, and nuance.
When MTPE is the right choice
Post-editing works particularly well in three scenarios.
High volume with repetitive language. Technical manuals, product datasheets, support documentation, and knowledge base content tend to repeat structures and terminology. Machine translation engines handle this type of text well, especially when supplied with translation memories and client glossaries. MTPE can reduce turnaround times significantly without compromising quality.
Internal or temporary-use content. Internal reports, inter-team communications, or content that will not be published publicly can benefit from light post-editing. The goal is comprehension, not publication quality.
Language pairs with strong machine translation output. English-Portuguese, English-Spanish, and English-French produce considerably better machine translation results than less common pairs. For languages with fewer digital resources, the quality of the automatic draft degrades, and the post-editing effort increases to the point where it can match or exceed the cost of human translation.
M21Global holds ISO 18587 certification, which defines quality requirements for the post-editing of machine translation. This means the MTPE processes the company applies meet internationally recognised standards, which matters to organisations that need to demonstrate compliance to clients or auditors.
When human translation is the only option
There are contexts in which MTPE is not appropriate, regardless of the post-editing level applied.
Legal and regulatory content. Contracts, terms and conditions, privacy policies, and regulatory filings require absolute precision and terminological consistency that go beyond what machine translation currently delivers. A meaning error in a contract can have direct legal consequences. Human translation by translators with legal subject-matter expertise is the only defensible option.
Marketing and brand communication. The tone, voice, and cultural resonance of marketing copy cannot be reproduced through post-editing alone. Localising advertising content, slogans, or conversion-oriented product pages requires a translator who understands the target market, not just the target language.
Medical and pharmaceutical content. Instructions for use, package leaflets, and safety data sheets have direct implications for patient safety. Errors are not simply quality issues: they are safety issues. Many markets require specialist human translation for this type of documentation by regulation.
Languages with limited machine translation resources. For projects involving mobile app localisation for markets such as Angola and Mozambique, where linguistic and cultural specificity is critical, human translation provides local relevance that machine translation cannot yet replicate consistently.
How to structure the decision
The choice between MTPE and human translation is rarely binary. Most medium-sized projects contain content that falls into both categories. A software product, for example, might have internal technical documentation that suits MTPE, alongside a user interface aimed at end users that requires human translation with close attention to context.
The determining factors are:
- Risk: what is the impact of an error? Incorrect information in an internal manual has different consequences from an error in a legal document or a safety instruction.
- Audience: is the text for internal use, or will it be read by customers, end users, or regulatory authorities?
- Language pair: does machine translation perform adequately for this pair and domain?
- Volume and recurrence: high-volume, repetitive content favours MTPE; unique or creative content favours human translation.
- Turnaround: MTPE delivers faster, but only when the automatic draft quality is high enough that extensive rewriting is not required.
For SaaS platforms and technology products, the decision has an additional layer: terminological consistency across versions, frequent update cycles, and the need to integrate with content management systems. These requirements are covered in detail in the article on ISO 17100 localisation for SaaS platforms.
How M21Global approaches this decision
M21Global works with both modalities and helps organisations identify the right approach for each content type. With ISO 17100:2015 certification for human translation and ISO 18587 certification for machine translation post-editing, the company provides documented quality assurance for both processes. The technology and software localisation services include content analysis, a recommendation on the most suitable process, and quality-controlled execution at every stage. For projects with mixed content types, both approaches can be combined within a single project, optimising cost and quality where each matters most. Contact M21Global to discuss your project and receive a clear recommendation on which approach fits your needs.
Related Services
Request a free software localisation quote
- Request a free software localisation quote
- Mobile App Localisation Angola Mozambique
- Iso 17100 Localization For Saas Platforms
- A Surprising Amount Of Content Available Online Is Generated By Artificial Intelligence
Frequently Asked Questions
What does MTPE mean in translation?
MTPE stands for Machine Translation Post-Editing. It is the process of reviewing and correcting a draft produced by a machine translation engine, so that the final output meets the quality requirements defined for the project.
Is machine translation post-editing cheaper than human translation?
Generally yes, but it depends on the quality of the machine-generated draft and the level of post-editing required. For language pairs with strong machine translation performance and repetitive content, MTPE can reduce costs significantly. For low-resource languages or complex content, the revision effort can match the cost of human translation.
What types of content should not be handled through MTPE?
Legal documentation, medical and pharmaceutical content, conversion-oriented marketing copy, and any document with regulatory or safety implications should be handled by specialist human translators. The risk associated with errors in these contexts outweighs any cost or speed advantage from MTPE.
What is ISO 18587 and why does it matter?
ISO 18587 is the international standard that defines quality requirements for the post-editing of machine translation. Its practical relevance is that it makes MTPE processes auditable and reproducible, which matters when an organisation needs to demonstrate compliance to clients or regulatory bodies.
Can MTPE and human translation be combined in the same project?
Yes, and it is a common practice in medium or large projects where different content types have different requirements. Internal technical documentation can be handled with MTPE, while the user interface or marketing content receives full human translation.
