Re:The Inuit agree on a common writing system
by jcranmer on Monday October 07, 2019 @04:11PM
(#21185427)
Part of this may come down to corpus availability. The EU requires that much of its official document production be translated in the official language of every member country (which would include Irish and Latvian). With essentially bilingual translation pairs, it's much easier to feed the data into machine translation; and the need to feed this translation machine for official translation also means it's relatively easy to contract out translation work if you need to.
I've noticed that in terms of language support in general. Of course, languages with Western and/or wealthy populations seem to have better support. In Elasticsearch, I can set up a field for Irish or Latvian or Basque analysis out of the box (each with ~1 million speakers or fewer). But Zulu or Uzbek or Malay (each with 10 million+ speakers), and sorry, no support.