Abstract
Within Relevance Theory, it has been suggested that extended metaphors might be processed differently relative to single metaphoric uses. While single metaphors are hypothesized to be understood via the creation of an ad hoc concept, extended metaphors have been claimed to require a switch to a secondary processing mode, which gives greater prominence to the literal meaning. Initial experimental evidence has supported a distinction by showing differences in reading times between single and extended metaphors. However, beyond potential differences in comprehension speed, Robyn Carston's 'lingering of the literal' account seems to predict qualitative differences in the interpretative mechanisms involved. In the present work, we test the hypothesis that during processing of extended metaphors, the mechanisms of enhancement and suppression of activation levels of literal-related features operate differently relative to single metaphors. We base our work on a study by Paula Rubio-Fernández, which showed that processing single metaphors involves suppressing features related exclusively to the literal meaning of the metaphoric vehicle after 1000 milliseconds of encountering the metaphor. Our goal was to investigate whether suppression is also involved in the comprehension of extended metaphors, or whether the 'lingering of the literal' leads to continued activation of literal-related features, as we take Carston's account to predict. We replicate existing results, in as much as we find that activation levels of literal-related features are reduced after 1000 milliseconds. Critically, we also show that the pattern of suppression does not hold for extended metaphors, for which literal-related features remain activated after 1000 milliseconds. We see our results as providing support for Carston's view that extended metaphor processing involves a prominent role of literal meaning, contributing towards explicating the links between theoretical predictions within Relevance Theory and online sentence processing.