•If the character between the opening brace [ and the first visible character is a space, then this often happens if you double-click on the inside part of a token and then copy that selection into another inside of a token (also by previously double-clicking on its inside part). In those cases, QuarkXPress inserts a blank space in front of the replaced text. This is a well-known QuarkXPress quirk, and you just have to be careful when copying the inside of tokens from one set of braces to another.
•Other possible characters (with ASCII values from 1 to 31) can be due to left-over encoding if you are using a Quark file that was previously used in a Quark-based semi-automated system. These systems often add invisible characters (usually ASCII code 4) in front of text that has a meaning. That is to say, instead of using a plain text token such as [1Dddd] to tell Quark that this is a weekday, these systems will leave the text Monday in place, but will insert an invisible character in front of it. If you are using files that were used in a previously used semi-automated solution, then you may want to clean up all the bits of text that will be converted to tokens, before performing the conversion.
In either case the token has no meaning and will be ignored until you clean up the token in the GridTemplate's QuarkXPress file.