... the word regex \W ...

It's important to understand what you're dealing with. The  \W (that's big-W) character class (see perlrecharclass) matches any character that is not a  \w (little-w) character. The  \w characters are sometimes called "word" characters, but IIRC they originate with the set of characters that are allowed in a C- or Perl-language identifier; that's why  _ (underscore) is included, but  - (hyphen), for instance, is not. So  \W is better described as the anti-word regex!

And I agree with davido's point here that if  [A-Za-z\W] really does the trick for you, then  [^_\d] is more clear, readable, maintainable, and IMHO preferable.

Update: Made "identifier" into a Wikipedia link.