We propose to address our specific hypotheses through a combination of (a) traditional reflective methods such as theoretically-informed pragmatic analysis, thought experiment and developed argumentation, and (b) controlled experimentation on adults’ processing of a range of kinds of metaphor and hyperbole, using both off-line and, crucially, on-line techniques. Here is an outline of the programme of planned empirical research:
1. To test our view that, while the understanding of both hyperbolic and metaphoric uses of words involves concept broadening, they differ in that the comprehension of metaphor (but not of hyperbole) also involves a domain shift, brought about by a process of concept narrowing, we will carry out the following two on-line experiments:
a. A cross-modal lexical priming experiment (a technique used extensively before by our postdoctoral RA, Rubio Fernandez (2005, 2007)) to test the activation levels of properties associated with the literal meaning of words used hyperbolically, on the one hand, and metaphorically, on the other. The prediction is that, across different interstimulus intervals (ISIs), we will get a different pattern of results for the two kinds of use. For example, while a hyperbolic use of the word marathon will continue to prime words associated with the literal meaning, like run and miles, at long ISIs, these will be suppressed at an earlier stage when the word is used metaphorically (as in ‘Writing a thesis is a marathon’) while words associated with the domain of psychological effort will be activated relatively early.
b. A measure of electrophysiological activity in the brain (ERP) during the processing of words used hyperbolically, metaphorically, or (apparently) both, to see whether there are significant differences across the distinct kinds of language use, as we would predict. This kind of evidence has been shown to be a potentially revealing complement to data yielded by behavioural measures (Van Berkum 2004, 2009) and has already been used successfully to test certain models of metaphor interpretation (Pynte et al. 1996; Coulson & Van Petten 2002). We are interested in the amplitude of the N400, an ERP component that can be used as an index of processing difficulty. It has been shown to be appreciably greater for metaphorical uses of words than for literal uses, but, as far as we know, it has not yet been enlisted during hyperbole comprehension.
2. To test the two process account of metaphor, we will carry out a large-scale study consisting of four sets of experiments each employing a different experimental technique so as to achieve a strong test of the theory and, it is to be hoped, converging evidence. The materials to be used in all four experimental settings will consist of three kinds: (i) extended metaphors (such as the one from Heller given above); (ii) single word metaphors in a literal context (which nevertheless biases for the intended interpretation); (iii) a wholly literal coherent discourse. In each set of triples, our target word (the same in all three cases) will occur some way into the passage, and will be used metaphorically in the first two conditions and literally in the third (the control condition), e.g. ‘boots’ in the extended metaphor from Heller above. The construction and pre-testing of these materials will be a major task in the first months of the project.
The four experimental paradigms are the following:
a. Off-line tasks in which participants are asked to produce a paraphrase (an interpretation) of the extended and individual metaphors, as well as to evaluate them for comprehensibility and aptness (see Glucksberg & Haught 2006).
b. On-line cross-modal lexical priming experiments in which the degree of priming by the target word (e.g. ‘boots’) of its literal associates (e.g. ‘shoes’) is measured at several temporal intervals from the offset of the word. Following earlier work by our RA, Rubio-Fernandez (2005, 2007), we are especially interested in the relatively long interstimulus interval of 1,000 milliseconds, as our theory predicts that it is at this point that the priming results for the target word in the extended metaphor (i) and the literal passage (iii) will converge, distinguishing it from the same word used as a single word metaphor (ad hoc concept) (ii).
c. Tracking of ERP waveforms during the processing of the target word in the three contexts. As noted in (1b) above, this kind of evidence has already been used successfully in comparing the processing of literally and metaphorically used words. Again, we are interested in the amplitude of the N400 component, which is highly sensitive to both local semantic and global discourse coherence. Focusing on the target word as above (e.g. ‘boots’), we will use the N400 to investigate whether integration of the metaphorically used word within an extended metaphor (i) more closely resembles integration in the literal control condition (iii), as we expect, or integration when it occurs as a single word metaphor (ad hoc concept) (ii).
d. Eye-tracking of participants during natural reading of the three kinds of stimuli. The various processing measures used in monitoring eye movements during reading (e.g. number and duration of eye fixations, regressive eye-movements (see Rayner 1998) should provide us with a third means of comparing the on-line processing of extended metaphors and single word metaphors in a discourse context, relative to literal uses of the same words. A tentative hypothesis is that, in processing extended metaphors, participants will fixate longer on particular metaphorically used words within the metaphorical scenario than on the single word metaphors (ad hoc concept cases) occurring in an otherwise literal context. More generally, we would expect a different pattern of fixations and regressions during the reading of the extended metaphors than for either of the other kinds of item. However, we are well aware that we need to sharpen our hypotheses here, which we will have time to do as this experiment will take place in the final year.
Page last modified on 06 apr 11 16:08 by Natalie Wilkins