When & How to Use Pre- & Post-Editing

Pollion Team

Technological advances have led to increased use of machine translation in the translation field. Machine translation (MT) was first developed in the 1950s to automate the translation process, without human involvement. While humans are still necessary for the translation process, there are some benefits to using machine translation.

Benefits of Using MT

These days, MT is increasingly utilized by many translation agencies—here are two main benefits of using machine translation:

  • Speeds up the translation process, making it faster to get your content to the target audience
  • MT is cheaper to use than human translators

However, humans are still needed to process machine-translated texts. This is where pre- and post-editing become important to the translation process. 

What is Pre-Editing?

Pre-editing is the process of editing the text before the machine translation. This may involve correcting errors in the source text (spelling, grammar, punctuation), while simplifying the text’s structure and removing any vagueness from the text. 

Pre-editing may also involve adapting the text so it’s a closer match to the texts used to train the engine. This can help the MT engine create a more accurate translation.

What is Post-Editing?

Once the machine translation is completed, it’s time for what is known as post-editing. Post-editing is the process of editing an automatic (machine) translation. The quality of machine translated text varies from project to project, depending on several issues:

  • Language pair used
  • The client’s determination of high vs low quality
  • Quality of source material

Once the automated translation is completed, it’s necessary to determine the desired output quality level. This can be done using one of two post-editing levels—light- or full-post editing.

Related Post: How to Translate a Document?

Light Post-Editing

This process takes the raw automatically translated text and makes as few changes as possible. Light post-editing ensures the translation is accurate and the grammar is correct. This process typically involves one or more of the following: 

  • Correcting typos, word and grammatical errors
  • Rewriting confusing sentences–partially or completely
  • Fixing machine-made errors
  • Deleting needless or redundant translation options made by the machine
  • Ensuring terminology is consistent, but with no in-depth term checking

A human translator is needed to ensure the MT text correctly conveys the ideas and concepts of the source. However, the resulting text doesn’t have to completely match the source text—the MT text needs only to convey the original concepts via translation. 

Light post-editing tends to be faster; only the major and critical errors are fixed, to ensure the reader can correctly understand and follow the text. No stylistic “polishing” is performed during the light post-editing process. The resulting text may sound a bit “robotic” or not quite right when it comes to tone and style; but the text is fluid and easy to understand.  

Full Post-Editing

Full post-editing work tends to be slower, cost more, and is more in-depth than light post-editing. The resulting translation must be completely accurate and stylistically correct, with correct terminology and tone. There should also be no errors in grammar. The translation should read naturally—as if it was originally written in the target language.

Tasks during the full post-editing process include all the tasks of light post-editing, along with the following:

  • All grammatical mistakes (typos, punctuation, spelling, etc.) are corrected
  • Correct format and tagging is applied
  • All cultural references are adapted, including idioms, examples, etc.
  • The translation should be perfect between the source and target text
  • Content should be consistent and fluent
  • Terminology should be checked against approved terminological resources, to ensure it’s consistent and appropriate.
  • Translations need to be cross-referenced with other resources

The final version of the translated text must be the same as if translated by a human—in all regards. The content must meet all the quality requirements as defined by the client. 

When is Post-Editing Used?

Post-editing is generally best for technical or scientific documents, where the number of word meanings are limited. The higher the technical complexity of the source text, the more specific the translation must be, in order for the machine to choose the correct words. 

Choosing Light or Full Post-Editing

How do you determine when to use light- or post-editing? There isn’t a simple answer, as the level of editing is often determined by the client and the type of project. Clients may have budget constraints and therefore expect the lower cost of light editing. Light editing is also faster, making it a better fit for the client’s turnaround needs. However, full post-editing may be required for more technical documents, which require a higher level of accuracy.

In the end, the quality level of any translation varies from project to project, and Pre-Editing is a crucial factor in ensuring the best possible outcome. The most critical part of the quality level is determined by the client’s needs and expectations. Each project needs to be thoroughly reviewed with the client to carefully define the project’s parameters, and whether or not machine translation and post-editing, combined with Pre-Editing, are used in the translation process.

Tags: translation