Six Tactics to Reduce Machine Translation Post-Editing
Click here to close
Click here to close
Subscribe here

Six Tactics to Reduce Machine Translation Post-Editing

Six Tactics to Reduce Machine Translation Post-Editing

Six Tactics to Reduce Machine Translation Post-Editing

It’s not uncommon for an enterprise with a maturing localization program to deploy Machine Translation (MT)—in fact, here at Moravia we believe that MT has a role in every large or growing globalization program.

And most MT programs involve some level of post-editing: the process by which a trained linguist works to bring the machine’s output to a level of quality agreed upon between the client and vendor. Their work takes care of quality, but how about the ubiquitous demands to boost turnaround time and reduce cost? You need to make sure their involvement meets all your expectations for ROI.

We talked to some experts inside Moravia and asked them for tips on improving the post-editing experience and effectiveness. Here are six things you can do to reduce, yet maximize, the amount of time spent on post-editing, while staying within the boundaries of schedule.

  1. Define the final quality. Quality is a moving target: if you ask five different people to judge the quality of a piece, you will probably get five different answers. Make sure all stakeholders agree on quality standards for each content type (web content, online help, user-generated content) and on the post-editing level required, then document them (usually as a framework based around the number and types of issues acceptable per a defined number of words).
  2. Write for localization. Tips for clean, clear writing include being consistent in terminology and sentence structure, using simple language, being succinct, and avoiding acronyms (where possible), abbreviations, idioms, and humor.
  3. Consider pre-editing. If the materials haven’t been written with localization in mind, or if they were written by a non-native speaker or a non-professional, then pre-editing can help. Editors revise against the same checklist as authors use when writing source content.
  4. Use Quality Assurance tools. Automated QA tools help post-editors run a final check on documents before delivery by flagging potential issues in the source or translation. QA tools can check spelling, terminology, punctuation, consistency, and style.
  5. Choose your post editors carefully. PE productivity is driven by the quality and experience of the resources used. Check for depth of experience: post-editing is a very different skillset than translation or linguistic review. Instead of converting source text to target text faithfully, a post-editor has to understand how an MT engine operates and what errors might be typical, and then fix all issues required to meet the requested quality bar. And lastly…
  6. Using Translation Memory technology, leverage as much material as possible before conducting MT. When you pre-translate text with past content, less has to be machine translated.

Every enterprise wants to get more content out to more markets more quickly. Using these suggestions, you can boost your post-editors’ productivity, optimize the time they spend, and increase the ROI of your MT investment.

 For more info:

  • Read about the types of post-editing—usually described as “light” or “full”—in this blog post.
  • Check out an interview with a post-editing expert here. There are some surprises in there!
  • Read up on the TAUS guidelines for post-editing.