is it just me, or do duolingo’s assumptions about the relative difficulties of going from target language audio to target language text vs going from native language text to target language text seem completely arbitrary?