All files contain data which can be interpreted as text. Even images, applications and sound.
The ascii extraction in File Juicer filters away most of the non-text portions of files, by removing very long words (if there are no spaces with reasonable intervals - some very long words are allowed (urls for example)), and there is few upper case letters. It also filters away binary data which is obviously unprintable characters.
The text extraction is loose when deciding what is text, and it will aprove too much as beeing text rather than stripping away real text.
If you drop a text file on file juicer it should produce an unchanged copy.
More about text extraction in the user manual.