MTPE or traditional translation? Good question.

For translation agencies like ours, the evolution of AI means we can offer a new service: machine translation post-editing (MTPE). This service involves generating a translation using an AI tool, followed by a thorough revision by our professional linguistic team, the result being a target text that combines machine precision with human insight and finesse. 

Just as a chainsaw shouldn’t replace a handsaw for all occasions, the post-editing of machine translations is no substitute for traditional translation. Instead, it is a distinct service that can deliver excellent results quickly under the right conditions. The first step in our process is, therefore, to determine the suitability of machine translation post-editing (MTPE) for a specific project. Several considerations factor into this decision, but in short:  the more a text deals with a personal subject matter, the more unique its style, and the more its content serves to persuade or engage a target audience, the less likely we are to recommend machine translation. In comparison, instructional or explanatory texts lend themselves well to machine translation. 

Cultural specificity is also an important factor. Since machine translation relies on reference texts, it tends to favour the majority language variants. In French, this means prioritizing European expressions and terminology. That’s why we don’t recommend machine translation for texts containing cultural connotations intended for a Quebec target audience, for example, as we would want to give slogans and advertisements a distinctly local flavour.

Finally, the quality of a machine translation largely depends on the volume of information to be processed. Generally, AI relies on reference texts, searching for matches in a database of hundreds of thousands of source and target texts. As a result, the longer the text, the more comparisons AI can make and refine its output. Machine translation is, therefore, best avoided for very short texts, such as social media posts, as the machine lacks context.

Once we have approved the use of machine translation for a project and AI has generated the translation, we don’t just correct typos in the translation. On the contrary, one of AI’s major advantages is that it generally doesn’t make typos. However, AI does have some major blind spots, such as: 

• Creative terminology (idiomatic expressions, slang, figures of speech)
• Double entendre/ambiguity
• Acronyms, titles and proper names
• Gender-inclusive language, abbreviated doublets
• Typographic rules and standards specific to the target language and its variants
• Tone and language register
• Errors in the source text

The role of our post-editing team, therefore, is to address these blind spots to ensure that the final text is accurate, culturally and contextually appropriate, and never misunderstood or misinterpreted. We also take the liberty of refining the style, where appropriate, to give the translation that touch of harmony, creativity and expertise that sets us apart.