Before filing a radar, I would like to get some feedback here: What is the Natural Language Processing language detection supposed to yield?
Take this input:
What der frohe Botschafter really means is the ‘bearer of Good News,’ with frohe Botschaft being the German translation of “the Gospel,” a.k.a. the Biblical Greek ευαγγελιον, which the Germans pronounce Evangelium, which is (aha!) where the name of their Protestant denominations comes from.
Of course, for "script", we get "latn" for every word except:
==============================
Script
==============================
Token: ευαγγελιον
↳ Grek
And except for this word, the dominant language is detected for every word:
==============================
Language
==============================
Token: What
↳ en
Token: der
↳ en
Token: frohe
↳ en
Token: Botschafter
↳ en
[…]
Token: ευαγγελιον
↳ el
Is NSLinguisticTagger supposed to recognize a different language in Latin Script? And if yes, in which circumstances does it? Or am I doing it wrong?