Liu, Pei and Connell, Louise and Lynott, Dermot (2018) Embodied-linguistic conceptual representations during metaphor processing. PhD thesis, Lancaster University.
2018liuphd.pdf - Published Version
Available under License Creative Commons Attribution-NoDerivs.
Download (31MB)
Abstract
Although metaphor processing has fascinated linguists and psychologists alike, the conceptual representations involved have not been fully examined. In the present thesis, I propose metaphor processing should be viewed as an aspect of language processing, involving conceptual representations that are both embodied and linguistic. The thesis includes five self-contained papers, which showed a detailed picture of conceptual representation that was flexible and dynamic. In the paper contained in Chapter 3, I proposed an operational definition of the effort to generate embodied simulation (i.e., the ease-of-simulation measure, or EoS). As a composite measure, EoS accounted for the speed of successful metaphor processing better than other rating tasks, which suggested that EoS could account for the underlying mechanism of metaphor processing, thus assumed to be embodied simulation. In papers reported in Chapters 4, 5 and 7, I studied influences of embodied simulation and linguistic distributional patterns on metaphor processing. These two components were both found to contribute to metaphor processing, and the interplay between them were were influenced by factors such as the depth of processing required and the time available for responses. Papers reported in Chapter 6 and 7 examined the EEG activations of embodied and linguistic components, in literal language processing and metaphor processing respectively. Both studies revealed that embodied and linguistic components performed various functions, each being activated at several time points. The linguistic component was activated first between 200ms-400ms after the stimulus onset, suggesting that it was involved in lexical and sublexical processing, which also supported the idea that it had a speed advantage compared to the embodied component. The latter was activated around 400ms, being responsible for semantic representations. Moreover, both components were activated again at the later stage of processing, indicating that both components were used and integrated for decision making.