Locating metaphor candidates in specialised corpora using raw frequency and key-word lists

When using corpora in metaphor studies, the question always arises as to how metaphors are to be located. Corpora are designed to facilitate the extraction of forms, but metaphors are not formally different from other words: they are merely words which are being used with a metaphorical sense. While it has been demonstrated that metaphorically-used words tend to collocate differently from their non-metaphorical counterparts, unless the distinct collocational patterns have been identified in advance, the researcher cannot take advantage of the differences in formulating his or her search queries. A further complication arises when the metaphorical meanings are not common enough to appear in a reference corpus, making it impossible to analyse their patternings. This chapter offers researchers a method for locating metaphors in corpus data which does not rely on prior investigation of word forms and their collocates, nor does it require extensive reading and annotation of the corpus texts. It intends to respond to the growing interest in domain-specific language in corpus studies and increasing use of corpora in metaphor research. It also responds to the pressing need for methods which are independent of dedicated software applications, which are not language-specific, and which can be used by the individual researcher using a PC concordance package to analyse an un-annotated corpus.