This sort of thing used to bite me in the begining as well. After deep thought on the subject (because of being bitten - and my pre-disposition at the time to other languages) I came to the following conclusion.

This is part of Perl's construct. In the Perl world - the difference is clear. Naming them similarly, and then using them interchangeablely or using them 'correctly' comes with time.

When I need an array - out of habit now - I do not create a scalar by that name (as in my mind it is now reserved for use when dealing with the array). The converse is true for me as well.

I personally, now that I am 'used' to this, find this feature of benefit as it lends itself to more streamlined coding and provides a built-in coding standard.