One thought on “The Widely-Spoken Languages We Still Can’t Translate Online

  1. tippiedog 28 Nov 18 / 9.09 am

    I have a Ph.D. in German and worked on machine translation projects for a few years back in the 90s. The big MT projects that started in the 1960s and 1970s were originally funded by the US government–CIA and DOD–and their language focus was based on US foreign policy priorities. One project that I worked on had worked on Farsi, Arabic and Vietnamese through the years.

    One interesting thing about this blog post is this seemingly innocuous statement, “Machine translation engines use parallel corpora to figure out regular correspondences between languages” Actually, for most of its lifespan, MT projects focused on translating languages ‘linguistically,’ but the big data approach (parallel corpora) swept it aside (well, current MT uses big data and some linguistic intelligence).


Comments are closed.