In the corpus-based approach, concordance data is extracted from a corpus on a random basis and examined within a colligational framework to generalize the collocational behaviour of a key word.