It's an implicit (and smart) count of occurrence.

Es.

BANANA (for kevin):

For i=1 we are sure that there is an occurrence for all this substrings that start with a vowel (s[i] = 'A'):

- A
- AN
- ANA
- ANAN
- ANANA
So we add len(s)-i = 6-1 = 5 to kevin score. 

For i=3:

- A
- AN
- ANA
(+3)

For i=5:

- A
(+1)