Services
Machine Translation Post-Editing into Portuguese
Machine translation post-editing has become a practical tool for LSP teams that work with large volumes and tight timelines. Modern MT engines produce output that can give translators a useful starting point for Brazilian and European Portuguese. This allows more content to be prepared quickly, while keeping final quality under human control.
MT output is not ready for publication on its own. It requires review, editing, and quality checks carried out by trained post-editors. Our teams handle this work for a range of content types and adapt the level of editing to what your client expects.
How we use MT alongside translation memories
Most MT workflows combine MT engines with translation memories. Pairs with approved translations come from the TM. Only the new or unmatched text is machine translated. This keeps terminology consistent and reduces unnecessary editing effort. MT is most effective when the source language is English and the target is Brazilian Portuguese, as many engines have strong data for this pair. European Portuguese also performs well, although it requires more careful review.
Types of post-editing
The right level depends on quality needs, audience, and how visible the text will be once published.
Full post-editing
Full post-editing aims for a final text that reads like a human translation. The output must be accurate, clear, and consistent in tone and terminology. This level suits content where quality expectations are high and errors cannot be tolerated. It also suits accounts where the brand voice or style guidelines require close attention.
Light post-editing
Light post-editing focuses on clarity. The text should be correct in meaning and free of confusing phrasing. Minor issues in tone, grammar, or style may remain. This level works for content where understanding matters more than perfect polish, or where readers accept small linguistic issues.
When MT may be used without review
Unedited MT is rare. It applies only to cases where quality is not important and the brand will not be judged by the result. Examples include forums where users expect imperfect language or internal resources that exist only for quick reference.
Why clarity on quality levels matters
MT workflows rely on clear expectations. Each client has a different tolerance for stylistic issues. Before we begin, we review the requirements with your team and define the level of quality that fits the account. This avoids confusion and helps ensure that the final result matches what the client expects.
Our MTPE approach
TextTrans has trained a dedicated group of linguists who specialize in machine translation post-editing. They understand how MT engines behave and how to correct issues without rewriting the entire text. Their work is supported by our quality process and the tools we use for consistency and accuracy.
Training provided
We train linguists in:
- How MT engines work
- Types of MT systems
- Differences between full and light post-editing
- Common MT issues and how to resolve them
- How to review MT output efficiently
- How to use reference assets during MTPE
For client-specific accounts, we also review:
- Sample content
- Expected quality levels
- Language assets and terminology
- Account challenges
- Review patterns appropriate for MTPE
Building and supporting MTPE teams
Linguists assigned to MTPE projects are selected based on experience, subject knowledge, and ability to work within the required quality level. They complete tests for full and light post-editing and receive feedback on their work. During the early stages, we review samples closely to confirm that quality meets expectations.
Our quality tools check for issues such as inconsistent translations, missing content, unmodified MT output, number mismatches, and terminology deviations. This keeps editing controlled and reduces the risk of overlooked errors.