PUB - Publikationen an der Universität Bielefeld

This paper examines patterns of metaphor in usage. Four samples of text ex-cerpts of on average 47,000 words each were taken from the British National Corpus and annotated for metaphor. The linguistic metaphor data were col-lected by five analysts on the basis of a highly explicit identification procedure that is a variant of the approach developed by the Pragglejaz Group (2007). Part of this paper is a report of the protocol and the reliability of the p rocedure. Data analysis shows that, on average, one in every seven and a half lexical units in the corpus is related to metaphor defined as a potential cross-domain mapping in conceptual structure. It also appears that the bulk of the expression of metaphor in discourse consists of non-signalled metaphorically used words, not similes. The distribution of metaphor-related words, finally, turns out to be quite variable between the four registers examined in this study: academic texts have 18.5%, news 16.4%, fiction 11.7%, and conversation 7.7%. The sys-tematic comparative investigation of these registers raises new questions about the relation between cognitive linguistic and other approaches to metaphor.