Post-Editing Machine Translation

Meet the demands of a changing market and future-proof your services

Machine Translation is a very cost effective way of translating large volumes of content. Automated translation is becoming more widely accepted and is used by many LSPs or agencies, as they strive to meet the demands of their customers by providing faster turnaround times. Used alone, automated translation is not a replacement for content that requires high quality human translation but with the intervention of the Post-Editor to edit, modify and correct the pre-translated text, machine translated text becomes a viable alternative service offering. With this the demand for professionally trained Post-Editors is rising.

Benefits of the Post-Editing Course

The scope of the Post-Editing Certification is to introduce the translator to the techniques and skills involved in post-editing Machine Translation (MT) output. It provides a theoretical overview of Machine Translation and post-editing, together with practical examples of best-practice post-editing (PE) and recurrent issues. 

The course is aimed at translators who want to start taking on post-editing work, or translators who already had contact with post-editing but want a more in-depth explanation on how MT engines are created and how to make best use of Machine Translation output. This course will enable translators to meet the demands of a changing market and will give them the core skills they need to future-proof their services and offerings.


Course Details

Module 1: History and Development of Machine Translation

A brief history of MT is presented, both general and SDL specific. Following on from the historical overview, the current situation of the international market is presented, backed up by figures on MT usage, internet users, cloud computing technologies and social networking tools. The number of end-users with different language requirements continues to grow and thus the demand for translation of content steadily increases. MT and post-editing are the only tools able to respond to these trends.

Module 2: Machine Translation Technologies and Engine Creation

This module focuses on the different MT technologies and discusses rules-based MT and Statistical Machine Translation (SMT). SMT “learns” to translate by analyzing large volumes of previously translated content. The more training data and the more consistency there is in this data, the more accurate the process becomes. SMT provides a more fluent translation with some context-sensitivity and better reflects the style of the training material. SMT is the technology of choice at SDL and the different engine types used at SDL are explained: Baselines, Verticals and Customizations. 

On the basis of the different engines, the process of creating statistical MT output is then presented: SMT output is created through TM cleaning, creating trainings and testing the chosen training. This section also includes an overview over the various automated evaluation mechanisms available.

Module 3: Post-Editing (understanding PE and effective PE) and use of SDL BeGlobal Baselines in SDL Studio

At this point, the guide begins to focus on post-editing. PE is a new phase that replaces conventional translation for MT projects. It is a skill that translators develop with time and experience – there is a learning curve associated with PE. The guide explores the differences between light post-editing (also referred to as post-editing to understandable quality) and post-editing to publishable quality. The market makes a distinction between these two levels of post-editing.

Post-editing to publishable level is the highest quality standard as final quality should be comparable to conventional translation. Post-editing to understandable quality, or light post-editing is normally required for low visibility text, or texts that would not otherwise be translated for a client as it would be too expensive and time-consuming.

This section also contains tips on how to post-edit effectively as well as common patterns to watch out for when post-editing statistical MT output.

Through SDL Trados Studio, everyone can have access to BeGlobal Baselines. It is possible to apply BeGlobal Machine Translation to the files received from clients and post-edit the output to full publishable quality. SDL BeGlobal Community can be added in a similar way to TMs in Studio.

Finally, the guide provides post-editing examples in different languages.

Certification Exam Details

The exam is accessed online and comprises 30 multiple choice questions based on the supporting materials. Students have 3 attempts to pass the exam and achieve certification status. 

Each exam will contain: 

  • 5 questions from module 1 
  • 10 questions from module 2 
  • 15 questions from module 3 

There is a total of 30 questions per exam with a pass mark of 20. 

The time allocated for finishing the exam will be 30 minutes: 

  • 5 mins for module 1 
  • 10 mins for module 2 
  • 15 mins for module 3 

At the end of the course, successful participants will receive Certification status and will be eligible to use the logo on their personal communications, via download from their SDL My Account.

Purchase video based training