Terralíngua combines machine translation with human post-editing to help organizations manage multilingual content. For suitable material, this approach can reduce production effort and cost while retaining professional language review. We also offer human translation, recommending a workflow based on the content, target languages, intended readers, and delivery requirements.
What AI Translation Does
Machine translation (MT) produces a translation automatically. AI translation tools can generate fluent text, but fluent wording does not establish that technical meaning, terminology, or instructions are correct. The output needs to be assessed against the source and the requirements of the assignment.
For example, a sentence in an equipment manual may read naturally while changing a condition or omitting a restriction. Repetitive product information may be a useful candidate for machine translation, while ambiguous instructions or creative messaging may require substantial intervention. The decision should consider the content and its intended use, not word count alone.
How Terralíngua Approaches Human Review
Post-editing means checking and correcting machine-translated content against the source. Our native-speaking linguists draw on subject knowledge, reference materials, and client terminology to address meaning, omissions, terminology, grammar, and style within the agreed scope. Questions in the original material may still require clarification from the client.
Our ISO 18587 certification covers full human post-editing of machine translation output. Our ISO 17100 workflow applies to human translation with revision. These are distinct services: limited editing of machine output is not equivalent to full post-editing under ISO 18587.
The review scope should be agreed before production begins, including the intended use, terminology requirements, and any client approval steps. For safety-sensitive or regulated material, the client retains responsibility for the required technical or regulatory approval; translation does not replace that review.
Choose a Workflow for Your Project
The following options involve different levels of intervention. Their suitability depends on the content, language pair, quality requirements, and intended use.
- Human translation with revision: A qualified translator produces the translation, followed by revision by another qualified linguist under the applicable ISO 17100 workflow. Translation memory and terminology tools can support this work without replacing the human translation step.
- Machine translation with full human post-editing: A qualified linguist checks and corrects machine output against the source to meet the agreed specifications. Our ISO 18587 workflow applies to full post-editing. Its suitability depends on the content, language pair, and quality of the initial output.
- Machine translation with limited human editing: The agreed scope focuses on selected corrections for a defined use. It should not be treated as equivalent to a fully post-edited, publication-ready translation. We discuss its limitations and suitability before agreeing to this approach.
Where Technology Can Help
- Production time: Machine translation can accelerate the initial language conversion. The overall schedule also depends on source preparation, the amount of correction required, formatting, and approval. Generating a draft quickly does not mean that the finished deliverable is ready.
- Project cost: Usable machine output can reduce the effort required to produce a translation. Poor output may require extensive rewriting, reducing or eliminating the expected saving. A representative source sample helps us assess the likely work before recommending an approach.
- Recurring content: Product information, documentation updates, and other repeated content may benefit from a combination of translation memory and machine translation. New or changed passages still need attention to their context and relationship with previously approved material.
- Translation memory: A translation memory stores source passages alongside their translations for reuse. It can help carry approved wording into later assignments and is useful in both human translation and post-editing. Storing a translation in a memory is a separate activity from training an AI model.
- Terminology: Client glossaries and reference materials help guide word choices and review. A preferred term may depend on the product, audience, or context, so automated suggestions still need to be checked against the project requirements.
- Confidentiality: The choice of technology must take account of your information-security requirements. Data retention, access, and any permitted use of content for model training should be clarified before confidential material is submitted to a translation system.
What Human Review Needs to Check
Review must go beyond improving how a sentence sounds. Depending on the material and agreed scope, checks need to address issues such as:
- Missing, added, or altered information, including conditions, warnings, and exceptions.
- Technical terms, product names, and abbreviations that need consistent treatment.
- Numbers, units, dates, and identifiers, with any conversion or adaptation agreed with the client.
- Meaning that depends on surrounding text, illustrations, interface elements, or the intended audience.
- Tone, regional language conventions, and wording that may be fluent but inappropriate for the intended use.
Where the translated content will appear in a document, interface, or video, language review also needs the relevant context. Layout checks, functional testing, and synchronization are separate production tasks that should be included in the scope where needed. Reviewing isolated sentences cannot establish that the finished material works as intended.
Plan Your Translation Project
Share your source files or a representative sample, target languages and regional variants, intended use, and required output formats. Include any terminology or reference materials, confidentiality requirements, and your preferred deadline. This helps us assess the work involved, recommend a review scope, and identify any formatting or production tasks.
Request a quote for your translation project.