This paper aims at reporting the development of a method for metaphor identification in computer corpora. The method was tested on a particular corpus, namely of investment conference calls, and comprises procedures that work from the bottom up, and rely on marked frequency, collocation and semantic similarity as signalling devices for metaphor. As such, the method is an example of corpus-driven research into metaphor. The application of these procedures yields a number of metaphor candidates, which are then checked manually through concordances.
Corpus Linguistics; metaphor; business; conference calls
