Post-Editing

Post-editing solutions to the largest MLVs, LSPs and global direct clients for their into Turkish (N)MT projects.

We have built an in-house team of linguistic resources trained in MT post-editing to provide you the post-editing of the content that was machine-translated.

TYPES OF POST-EDITING

Collaboration between post-editors and machine translation

The quality of machine translation output depends on many parameters including:

  • Language pair
  • The type of MT system (rule-based, statistical, neural, etc.)
  • The availability and size of domain-specific translation memories
  • Quality of the translation memories
  • Text type to be translated

We will provide you different services following the MT Post-editing Guidelines of TAUS (Translation Automation User Society):

FULL POST-EDITING

High quality translation at the level of professional human translation. It is also generally referred to as “publishable quality”. And it is the service that we recommend.

LIGHT  POST-EDITING

A comprehensible and accurate translation quality but stylistically it is not as good as human translation. This quality level is also called “good enough” quality.

Light post-editing is recommended in cases in which;

• the machine translation output has a sufficient quality
• the text type to be translated is “highly perishable”, which means that the texts will not have a long lifespan, the project has a large volume of content but the translation budget is limited.

 

Our MT post-editing team leader Gokhan Dogru is a PhD candidate in Translation and Intercultural  Studies, Autonomous University of Barcelona and a member of the Tradumatica Researcher Group. His research focuses on “Turkish><English Statistical Machine Translation and Postediting”.

TURKISH POST-EDITING

Machine Translation from and into Turkish has entered into the agenda of language service providers very recently thanks to the advances provided by neural machine translation.

In 2016, when Google Inc. announced their success with neural machine translation (NMT), one of the first languages to which they implemented NMT was Turkish (out of 103 languages that they support).

This advance has become a milestone for post-editing from and into Turkish. Since we have been following closely the developments in the machine translation area through our MT consultants, we have quickly adopted a strategy for both creating custom NMT systems and post-editing guidelines specific to Turkish language.

Besides, we have integrated these systems and guidelines into our translation workflow so that we can deliver our post-editing services in a speed compatible with that of the machine translation itself.

Localex was founded in Istanbul, Turkey. We offer price-competitive, customer-focused, and quality-controlled services with our 10 in-house best-choice professional linguists, and have the capacity to handle large volume projects in-house at tight deadlines. Each month, we translate over one million words from English, German, French, Spanish, Italian, Arabic, and Farsi into Turkish for the largest MLVs, LSPs, and global direct clients in a wide variety of industries. We specialize in technology, marketing, life sciences, industrial manufacturing, and legal texts.