#agglutinative-languages

Question 1: Single choice

Turkish builds long words by chaining suffixes onto a stem: evlerinizden ("from your houses") is ev "house" followed by a chain of suffixes (ev-ler-i-n-iz-den). Which kind of tokenizer suits such a language best?

Subword, so frequent stems and suffixes become tokens and rare words are built from them1
Word-level, so that each inflected form gets its own vocabulary entry2
Character-level, since one token per letter loses no information at all3
Whitespace splitting followed by lowercasing, which removes the variation between word forms4

#agglutinative-languages

Timed test