The core of our bespoke human translation process and neuro machine-translation process is made up of 4 essential steps, including two internal QA steps, which are a key differentiator in the localisation industry, whereby our competition presents a process of 2 or 3 steps only. The sequence of tasks from our standard translation process unfolds as follows:
Translation or neuro machine-translation
The source text is translated into the target language by a professional native translator, expert in the industry the content handled derives from. In the case of neuro machine translation, our team prep the translation engine and run it against the content in scope.
This is our first internal QA step following the translation step. Once the translator has completed the translation, our project team will run the first mandatory checks to see if the translation is compliant with the project requirements (spellcheck, adherence to glossary, punctuation, empty segments, numeral matches, translation consistency, etc). This allows us to block the translation at this stage if we spot issues to allow the translator to rework some parts and fix the oversights our team detect. Our team will then verify that improvements have been made before carrying on to the next stage of our translation workflow. In the case of neuro machine-translation, our team run checks on the successful application of the translation memories, adherence to the glossary, numeral equivalence, potential empty segments, etc. All potential problematic instances our team run into will be flagged to the post-editor.
Revision, also called proofreading
The target translation is checked against the source text by a professional native revisor or proofreader. In the case of a translation generated by neuro machine-translation, our post-editor will fix all potential oversights left by the machine and re-establish the natural flow of the target language.
This is our second internal QA step following the revision step. Once the proofreader has completed the revision, our project team will run the final batch of QA checks (spellcheck, adherence to glossary, punctuation, empty segments, numeral matches, translation consistency, etc). Our team has the opportunity here to block the proofread translation and to liaise with the proofreader or post-editor to clarify the last doubts and ask for some amendments if issues have been flagged during the QA checks performed at this stage. If no issues, our project team validates the translations and move them forward onto the next steps of the project workflow.
The verification and validation steps empower our project managers and allow them to have their say in the translation process. Our team is well trained to run QA checks and notice potential issues. These two steps offer our project managers room to discuss the work delivered by our linguists immediately on the spot after receiving the translated copies. This instantaneity is important when discussing translation and giving feedback on it as we want to strike while the iron is hot, meaning that the content is still fresh in our linguists’ mind and fixes can be done swiftly.