While language contains millions of words, most units will only know between 10-30 thousand. Some will only know a few hundred.
Certain words will never appear in a unit's initial rap sheet, such as swear words or words denoting a religion outside of Scripture, such as "god". Those rare few Erfworlders who are aware of and try to use them feel pain, as if the Titans consider speaking those words a crime.
Signamancers are capable of feeling out words outside of their normal rap sheets, including words from outside their world.