thank you

unless problem was fixed check your code for typos. I've tried both ways and they did work.

the problem is that in my native language the sum of a sequence of integers means a bit different thing. for instance the sum of a sequence of 2 6 2 is 10, not 12 or 38 or 66 etc.

I'm not sure how to explain it clearer than the description itself (do the worked examples not help?). Tackling this a different way, can you explain what bit of the description you don't understand?

That's a problem with how you expressed something in your code, not a problem with how the description is expressed.
Dropping starts happening when there are more than n of a value, not when there exactly n.

That is because when you use the ascii value (i.e., starting from 97+) the length of the word plays a bigger role than it should. Every extra letter would be a minimal addition of 97 instead of a minimal addition of 1. For instance, it would cause 'aaaa' (4*97=388) to score higher than 'zzz' (3*122=366). Of course, when you use the actual alphabet index of the letters, 'aaaa' (4*1=4) would score way lower than 'zzz' (3*26=78).

Why? If there is some language where the behaviour is not as stated in the kata description just say that, changing the title would be wrong.

Might depend on the programming language. It isn't all that hard once you know you need to converr it.