I entirely agree with tafoxx that the description is not clear enough for someone to know how to handle special characters correctly. Consequently many people fail the test with random input (as I did), and it is difficult to determine why without seeing the exact input of the test case.

Changing the first item under 'Assumptions' to read as follows would greatly improve the specification:

• A word is a string of letters (A to Z) optionally containing one or more apostrophes (') in ASCII. Non-word characters (e.g. #, %, !, &) should be treated as whitespace.