The Role of Literal Features During Processing of Novel Verbal Metaphors

An ERP study on the late stage of Chinese metaphor processing

Psycholinguistic models of metaphor processing remain a subject of debate. A prime-probe design using Chinese materials with a specific time span (300 ms) was applied to test the mechanisms of metaphor processing. Conventional and familiarized metaphors were designed as primes, followed by a probe word semantically related to the prime metaphor (MT), a probe word related to the literal meaning of the final word of the prime metaphor (LT), control/unrelated probe word (UT), or non-word. Event-related potentials (ERPs) elicited by the probes were recorded to examine metaphor processing. In N400, results revealed that UT and LT elicited significantly more negative waveforms than MT in both primes. MTs and LTs showed no difference between conventional and familiarized metaphors, suggesting that metaphorical meaning may be accessed directly, regardless of whether conventional or familiarized metaphors. The results were generally compatible with the direct processing model.

Suppression of literal meaning in single and extended metaphors

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.

N400 differences between physical and mental metaphors: The role of Theories of Mind

Whether Theory of Mind (ToM) contributes to metaphor understanding has been largely investigated in language acquisition and decay. Yet we know very little about the role of ToM in real-time processing of metaphors in neurotypical adults. Here, we tested the relationship between ToM and metaphor through Event Related Potentials (ERPs) by capitalizing on the difference between metaphors inviting inferences on physical (Boxers are pandas) vs. mental aspects (Teachers are books). Physical metaphors involved a larger and sustained negativity compared to mental ones. This pattern resembled concreteness effects and suggests that physical metaphors may benefit from both verbal and perceptual information. Moreover, higher scores in the Reading the Mind in the Eyes Test (RMET), but not in the Animation task, were associated with a reduction of the N400 amplitude for both physical and mental metaphors. When exploring the ERP temporal trajectory with Generalized Additive Mixed Modeling, earlier differences between metaphors characterized individuals with higher RMET scores. Among the various ToM components, thus, emotion recognition seems to be involved in the processing of metaphors in general, with an earlier impact on the mental type. These findings highlight the multifaceted nature of metaphor, at the crossroad of language, social and perceptual experience.