Tony BERBER SARDINHA
(Catholic University of São Paulo, Brazil)

ABSTRACT: In this paper, I take a look at the extent to which the distribution of lexical metaphor matches the probability patterns proposed by Halliday. The research was motivated by the fact that lexical metaphor has long been ignored in SFL, and so there was no indication that the use of lexical metaphor would be predicted by patterns of distribution that were originally meant to refer to typical systemic categories (such as polarity). The research was based on the analysis of two corpora, one of conference calls and another of general language, both of Portuguese. The conference calls were read one by one and all of its metaphors were coded; then the same metaphorical lexis was searched for in the general language corpus and analyzed. After that, the frequency of metaphor use was calculated for each corpus, its distribution was described in both corpora, and this was compared to Halliday’s original proposal. Finally, a comparison was carried out between the two corpora to see to what extent there was ‘resetting’, or shifts in probability patterns, as predicted by Halliday. The results indicated that metaphor use largely matched Halliday’s probability profiles, and there was evidence of resetting, with the exception of cases where ‘zero resetting’ occurred. The paper closes with remarks arguing for Systemic Functional Linguistics to stop ignoring lexical metaphor, as it is a very important device for linguistic expression as well as link to conceptual structures in the mind.