ABSTRACT
Corpus annotation, whether lexical, morphological, syntactic, semantic, or any other, brings additional linguistic information as an added value to a corpus. The annotation scenario might differ considerably among corpora, but it is always based on some formalism that represents the desired level and area of linguistic interpretation of the corpus. From the simple annotation of part-of-speech categories to the shallow syntactic annotation to semantic role labeling to the “deep,” complex annotation of semantic and discourse relations, there is usually some more or less sound linguistic theory behind the design of the representation used, or at least certain principles common to several such theories.
