Corpus annotation refers to the process of marking up one's corpus with linguistic information about the contents of the corpus. This is often used to add part of speech and syntactic information to the corpus. Different TagSets are used for this purpose.
I specify that the information is linguistic, because there is also non-linguistic information that can be added; for example, one might add dates, information about where the corpus data came from, or even formatting information, etc. I believe that this is usually called corpus markup, rather than annotation.