The ASCII question mark is typically what you get when something tries to convert some unicode character into some non-unicode character set that does not contain the character in question.

For example, the following script will produce "foo??", because the string literal has unicode Cyrillic for the fourth and fifth characters, but perl is being told to convert it to iso-8859-1 (Latin-1), which does not contain any Cyrillic characters -- that is, the unicode code points for Cyrillic cannot be mapped into the single-byte character codes for Latin-1, so the conversion produces "?" instead.

perl -MEncode -le '$_=encode("iso-8859-1","foo\x{041d}\x{0418}"); prin
+t'
[download]

It's not just perl that does this. Anything/everything that supports conversion between unicode and other encodings will behave the same way when faced with the same inappropriate task.

To figure out where the question marks are coming from, figure out the last point where the data were in unicode, and what sort of bad assumption is being made at that point to convert the encoding to something else.