The development of artificial intelligence is not only changing industrial processes but also the translation workflow as we know it. In the localization industry, the use of powerful engines to produce machine translation (MT) output is becoming more frequent. Because of this, the role of the translator and other professionals in the industry is being redefined. MT cultivates new roles and needs while other tasks are no longer required. The most relevant evolution created through MT involves the translator. This new role isn’t in charge of the translation anymore but of the edition of the output generated by the engine. This process of revision is called post-editing.
What exactly is post-editing?
When we examine post-edition, we are referring to the process of improving a MT output. The textual product is modified with two main goals. The first is to enhance that particular text in order to get a readable and understandable output. The second is to improve the MT engine with the linguist’s work and feedback. This process involves the job of localization or language engineers. It is important to note that post-edition requires specific skills. A post-editor should be very aware of details to detect errors and make pertinent corrections. Also, he or she should be efficient, since in this task time is everything. The goal of post-editing is to get a correct text with very quick and short modifications.
Furthermore, when dealing with MT projects, it is important to know which text types are a better option for this technology. Technical or scientific texts are more suitable for MT and post-editing than marketing materials, video-games or audiovisual products. This is because these are related to other media (e.g., video, audio) and have more creative content as well as higher complex sentences. Such factors result in harder input for MT.
Types of post-editing processes
There are two different processes of post-editing. First, light post-editing consists of implementing quickly a small number of changes so the MT output is considered acceptable. The expected corrections are the following:
- Omissions and additions
The text can contain grammar or punctuation mistakes. The purpose of the light post-editing process is to obtain an acceptable and understandable text. Also to make information available to readers, not an output with human translation quality.
On the other hand, in the deep post-editing process more changes are expected since the quality of the output should be equivalent to the quality of a human translation. As a result, linguists involved in the task spend more time working on the MT product. In a deep post-editing project, the editor should implement light-editing changes, plus the following:
After the edition, the text will be suitable for publication and distribution.
Types and needs
Knowing the difference between both post-editing types is key for a successful project. The process to be applied always depends on the clients’ needs. That way they can truly take advantage of MT technology.
- Post-editing Highlights: What to Correct
- Linguist Profiling: What Makes an Ideal Candidate for Post-editor
- Lead Linguist Bibiana Cirera’s View on Machine Translation
- 4 Stages and 8 Rules for Successful Post-editing