摘要：Computer based corpora which have been statistically analyzed have lead to the recognition of frequent lexical sequences (also known as multiword expressions, prefabricated chunksor collocations¨Cin a loose use of the term¨Cas a very important component of language production. Spontaneous interactions demand the storage of the phrasal element during the encoding and decoding of the message. Corpus studies, apart from improving the descriptions of lexis contribute, by means of the Applied Cognitive Linguistics proposals, to the learning process of these units with activities conveniently designed. This article brings an approach which combines the applications of corpora with the benefits of collocations ¨Cin a restricted use of the term¨Cwhen the lexical competence is being developed. The aim is to provide the ability to identify collocations by using the input contained in the corpora, which it will lead to the production of many different combinations, once the connections between a predicate and its arguments have been established
