That's why ikegami suggested using a hash of arrays. If you push a new tag ID onto the array you will be able to track all occurrences without overwriting existing data. Your example would look something like this:

%text = (
    'He' => [ 1, 2 ],
    ...
[download]

If, on the other hand, you were using a straight hash:

%text = (
    'He' => 1,
    ...
[download]

then yes, the first tag would be replaced by the second.

You may find perldsc and our Tutorials section (specifically Data Types and Variables) helpful.