Read on Twitter
Machine translation for websites: Pros and Cons? #webinar @weglot
(1) Since 2016 and neural translation, language is statistically analyzed, a word is translated according to a context (sentence) + the existing web contents.
DeepL does 11 languages, Google translate does 109
(1) Since 2016 and neural translation, language is statistically analyzed, a word is translated according to a context (sentence) + the existing web contents.
DeepL does 11 languages, Google translate does 109
(2) To localize a website, depending on the audience, industry/content/language, the quality expected, you can choose to use
100% machine translation: quick & sometimes dirty
machine translation post-edited = hybrid solution
human transl : good quality/slow
transcreation
100% machine translation: quick & sometimes dirtymachine translation post-edited = hybrid solution human transl : good quality/slowtranscreation
(3) Post-edited machine translation became dominant in 2020 (Memsource data): it is very much faster and working well in some cases.
But not working well for marketing & creative content where culture is key: landing pages, main product page, social media posts, payment... #MTPE
But not working well for marketing & creative content where culture is key: landing pages, main product page, social media posts, payment... #MTPE
(4) The website infrastructure should be localized very clearly, key parts of a website cannot be translated by a machine (no SEO possible on a non-optimized page).
(5)
Should a freelance translator provide post-edition? Yes
It is always better than not selling at all!
Translators need training to use and "train" (=improve) engines for their clients.
Should a freelance translator provide post-edition? Yes
It is always better than not selling at all!
Translators need training to use and "train" (=improve) engines for their clients.
(6) On main translators' CAT tools, you can use engines already.
Memsource for instance has added several neural machine translation engines: Google, Amazon Translate and a translator can see the results from both. (for free?)
Memsource for instance has added several neural machine translation engines: Google, Amazon Translate and a translator can see the results from both. (for free?)
