Over the last decade, the concept of Post-Editing emerged in the market of translation. It began to be used more frequently once the automated translations entered the field slowly, but steadily. 

There are a number of definitions to explain the term Post-Editing. It is often confused with editing and many translators and editors interpret differently the notion of Post-Editing. Post-editing is defined to be the machine translation output edited, revised by the human linguist. The Editing is done by a professional translator, the output of which is processed by the human mind while in case of Post-Editing, the output is generated by the automated system. Post-editing does not necessarily mean the light editing. Reviewing an automated translation requires high concentration. The sentences generated by the computer seem to be fluent and connected though a concentrated look may reveal several semantic and grammatical mistakes.

Post-editing involves deep editing to reach the level of the text to be completely published.    It must respect all punctuation, terminology, spelling and stylistic rules. In more detail, after post-editing, the translation should:   • reflect the exact image of the source text,   • include correct punctuation and spelling,   • syntactically reflect the connotations of the original language,   • include cultural changes (date and time formats, units of measurement, number formats, currency, etc. must be adapted).   • follow to the original language style,   • Comply with the end user's requirements as a reader.

How did the need for post-editing arise, and why don't translators like it?

The post-editing went through interesting stages. This was first mentioned in 1997, when professional linguists and translators post-edited the translation output generated by the computer. There was an explosion of content a part of which was based on the media and social networking as business drivers. The business world began to look at translation from a completely new perspective. The priority for enterprises was to provide their business content in different languages to a global audience. There were not enough translators to process this content, at the same time the translators working cycle was changing. The need for automated translation and localization were increasing. 

In 2016, Google announced the development of the GNMT, a Google Neural Machine Translation System. There was a great need for translators and specialist linguists who would do the post-editing work. In 2016-2017 alone, more data was created than in the whole of human history before. 

Before reaching high quality, automated translations had already gone through three main stages, which were based on the following 3 types of machine translations:

1․ Rule-based translations applying linguistic rules combined with dictionaries. The sentence was subjected to parsing of the source language, analyzing of the structure, converting to a machine-readable code and transforming it into the target language. The system used linguistic rules and combined dictionaries. The dictionary itself could have consisted of original language words, sentences, their translations, and detailed grammatical information.

2․ Automated statistical translations that generated translations using a large volume of previously translated content.

3․ Neural Machine Translation. It is considered to be the largest framework of Artificial Intelligence (AI).

It is estimated that around 800 billion words per day are translated through global public machine translation portals. Automated translations and Neural Machine Translations brought with them the need to re-edit translations made by the system. The linguist's task was to optimize the corpus data to ensure high quality machine translation. Working with engineers and other professionals, computational linguists were working on the language from more profitable viewpoint by analyzing the data and releasing high quality translation.

Due to the recent years demand, post-editing work has become a form of alternative translation. Translation degrees take several years of study and dedication. Translators spending many years on language experience and language aptitude, are not easy to accept the post-editing work. They reject their participation in machine translations as they realize the further impact on the translation market demand. However, new solutions are forcing them to rethink their professional approach and consider the PE as an alternative translation offer. The translator's profile is evolving to keep up with new translation and technology developments. Their work-cycle is not limited to translations. They are required to work on CAT tools or client owned tools and platforms. They carry out large-scale editing, revision, terminology and other non-linguistic work, where automation systems and translation service are harmoniously moving forward, posing new challenges for both translators and humanity in general.