AI translation with human review (commonly referred to as Machine Translation Post-Editing (MTPE) in the ISO 18587:2017 standard) is a powerful workflow that can expedite turnaround times and manage reduced budgets without sacrificing quality. This integration of advanced technology reflects the translation industry's commitment to meeting the high standards of today's global market while improving efficiency and effectiveness.

The most recent advancements in AI have further enhanced translation workflow opportunities, making them more accessible than ever before. When we talk about AI translation, we refer to the use of computer neural networks or AI software to produce automatic translations. While no AI system can match the precision of human translation, integrating human editors ensures that the final quality is comparable to that of a human translator. Maximizing the benefits of AI translation with human review involves strategic planning, such as launching a successful AI plan, which can significantly improve translation quality and efficiency.

 

Ideal content for AI translation 

Due to these advancements, AI translation with human review is an excellent choice for many types of materials, as detailed in one of our recent discussions on this topic, including:  

  • Employee communications 
  • External communications 
  • Technical manuals 
  • Datasheets 
  • Medical manuals 
  • IEPs  
  • ICFs 
  • Legal documents 
  • Financial documents 
  • Market research 

With those examples in mind, let’s discuss the quality levels outlined in the ISO 18587:2017 MTPE standard. Within the AI translation with human review workflow, there are two distinct levels: full post-editing and light post-editing. Both involve using AI-generated translations as a first pass, followed by a human editor, but they differ in their goals and final outputs.

 

Full post-editing 

Full post-editing represents the most comprehensive review of AI output and is explicitly defined by the ISO 18587:2017 MTPE standard. The requirements for post-editors at this level are similar to those for translating from scratch. The final output must be accurate, understandable, and grammatically and syntactically correct.

Post-editors focus on: 

  • Ensuring no information is added or omitted 
  • Editing inappropriate or incorrect content 
  • Restructuring sentences to ensure proper meaning 
  • Ensuring grammatical correctness 
  • Verifying terminology is correct and consistent 
  • Correcting spelling, punctuation, and hyphenation errors 
  • Adhering to client style guidelines, if provided 
  • Ensuring proper formatting, if required 

The thorough review by the post-editor ensures that the quality of the AI translation output is on par with a purely human translation approach.

 

Light post-editing 

Light post-editing, as the name suggests, involves a lighter edit than full post-editing. This level aims to use as much of the AI output as possible while making essential corrections. The focus here is on:

  • Correcting added or omitted information 
  • Addressing inappropriate content 
  • Clarifying incorrect or unclear information 

While full post-editing strives to achieve the quality of human translation, light post-editing has a different objective. The editing process for light post-editing is significantly simplified, with no emphasis on terminology, consistency, style, or formatting—preserving as much of the original AI-generated content as possible.

 

Which level do you need? 

Choosing between full and light post-editing depends on a few factors: 

Audience

  • Is this an internal document with limited viewership? 
  • Is this for research purposes? 
  • Is this a customer-facing document? 

Goal

  • Do you have specific terminology that must be adhered to? 
  • Does turnaround time supersede human-equivalent quality? 
  • Do you have a large amount of content with a limited budget? 

Light post-editing is faster and more cost-effective than full post-editing due to its reduced human involvement. It is suitable for large volumes of material with limited funds, research purposes, or internal documents with a small audience. Conversely, full post-editing is ideal for content requiring specific terminology, consistency, overall quality, or documents intended for a large, customer-facing audience. 

 

Final thoughts 

Regardless of the level chosen, it is crucial that your translation provider communicates the steps and workflows used to meet quality expectations. If you are unsure whether you are receiving light or full post-editing, just ask!  

By understanding the distinctions and applications of full and light post-editing, you can make informed decisions that align with your needs and ensure the highest quality outcomes for your translations. To explore how AI translation with human review can further enhance your workflow and meet your specific needs, learn more about AI-driven translation services below.

Related Blog Articles

When and When Not To Use Google Translate
When and When Not To Use Google Translate
Read article ›
Everything You Need to Know About AI Translation
Everything You Need to Know About AI Translation
Read article ›
When AI Translation Tools Put Your Organization at Risk
When AI Translation Tools Put Your Organization at Risk
Read article ›
Pros and Cons of Machine Translation
Pros and Cons of Machine Translation
Read article ›
images