Here's a nice rule of thumb when splitting up a string into components: If it's easier to write what you want, use m/what_you_want/g. If it's easier to write what you don't want, use split /what_you_dont_want/.

In your case, it's easier to write a regex for what you want (a word) than what you don't want (all between-word sequences):

my @words = $content =~ m/([A-Za-z]+(?:\'[A-Za-z]+)?)/g
[download]

This matches alphabetic characters followed by an optional apostrophe + alphabetics. This is obviously preliminary. Adjust as necessary according to your definition of a "word" ..