You might have become a victim of smart quotes, which is a feature of some programs (in particular word processors) to automatically replace regular ASCII single/double quotes with their curly counterparts, that typographers and designers are so fond of, because the quotes' opening (left-side) and closing (right-side) representations possess the slightly different look, as used in professional typesetting. This means that if you type: 'word', you'll get ‘word’ (or even some other form, depending on the locale), or "word" —> “word” (zoom in if you can't see a difference...) They are entirely different (non-ASCII) characters.

How to get rid of them depends on how they're encoded. In Unicode, they are the codepoints U+2018 - U+201B, while for example in CP1252, they are 0x91, 0x92, 0x82 (curly single quotes) and 0x93, 0x94, 0x84 (curly double quotes).

You can replace them using Perl's tr/// or s///, e.g.

# unicode
tr/\x{2018}-\x{201B}/'/;
tr/\x{201C}-\x{201F}/"/;

# CP1252
tr/\x91\x92\x82/'/;
tr/\x93\x94\x84/"/;

# or, if you have UTF-8 data which isn't properly flagged as such,
# you can try to directly replace the multi-byte sequence, as
# they're encoded in UTF-8, e.g.
s/\xe2\x80\x98/'/g;  # one of the single quotes,
# ...                # and similarly for the 7 others...
[download]

(This is just a guess, I can't tell for sure whether that really is your problem... you haven't told in detail in what way the quotes appear wrong...)