I agree, my first thought was to include non-english characters.

I like this solution. However the alphabet count > 1. So Character::isAlphabetic is not reliable way to filter distinct alphabet characters. Let say sentence has one or more letters from Greece of French. This solution gives wrong result in such case.