Actually use this:
{sum(
(!is_empty(QQ_SQ001)),
(!is_empty(QQ_SQ002) and QQ_SQ002!=QQ_SQ001),
(!is_empty(QQ_SQ003) and QQ_SQ003!=QQ_SQ001 and QQ_SQ003!=QQ_SQ002),
(!is_empty(QQ_SQ004) and QQ_SQ004!=QQ_SQ001 and QQ_SQ004!=QQ_SQ002 and QQ_SQ004!=QQ_SQ003),
0)}

Thomas : another good idea ?

If not : If you think it can be intersting reassign it to me.

Denis-

Belatedly, there is already a unique() function in EM. Per the description, "Returns true if all non-empty responses are unique"

So, you can mix this with the array column and/or row selection syntax in self or that to determine whether all non-empty values in that row or column are unique.

The attached group demonstrates three ways you can do this. I had hoped that the sub-question relevance would color code the rows and/or columns with problems, but they just turn everything pink.

So, you can close this ticket -- the functionality is already supported.

But , il already used unique(self) in ranking ! Why I put this one then ????

Oups, i think it's most for a really array unique system, (Removes duplicate values from an array).
Close this one ...

Have it :

I want to show all text checked in a multi question list, but with and, not ,.

list(MULTI.shown) show Sub-Question1,Sub-question3

concat(" and ",MULTI.shown) show Sub-Question1 and and Sub-question3

Something like array_filter
concat(" and ",array_filter(MULTI.shown)) can do the trick

OR:
concat(" and ",array_filter(MULTI.shown,trim)) (allow calable function, not only false