In this paper, we show that accessibility of syntactic information eases collocation extraction from corpora, and supports identification of lexical and structural restrictions related to collocations. For collocation identification we use a corpus that is automatically annotated applying a part-of-speech tagger and a phrase chunker.