Literary halo: ERP correlates of poetic metaphor

Introduction. Most of the existing neurophysiological literature on metaphors comprehension focuses on the processing conventional metaphors (Coulson & Van Petten 2002) or generic novel metaphors (Tartter et al. 2002, Lai et al. 2009). Except for a few studies including poetic metaphors in Hebrew (Arzouan et al. 2007, Faust & Mashal 2007), no extensive neurophysiological investigation on the processing of poetic metaphors has been carried out. Important theoretical insights have recently been offered by Relevance Theory in the framework of a general deflationary account of metaphor (Sperber & Wilson 2008): while not deviating from a normal use of language, poetic metaphors are likely to be more taxing due to the mass of weak implicatures involved in their comprehension (Pilkington 2000). Starting from the hypothesis that the literary feature has a crucial impact on processing, this study explores the cognitive costs associated with poetic metaphor comprehension through an ERP experiment. Based on previous literature (Coulson & Van Petten 2002, Arzouan 2007), a two stage process is expected, i.e., a stage of semantic access (around 400 ms) and a following stage of pragmatic enrichment (around 600 ms). In addition, we hypothesized that the costs related to the array of weak implicatures postulated by RT for literary metaphors should be reflected in late ERP components. Materials and methods. Stimuli consisted of 104 Italian poetic metaphorical phrases of the form " A of B " , taken from poems and novels, and an equal number of literal and semantically anomalous phrases (used as control condition): Metaphor: Somersaults of smoke / Literal: Prohibitions of smoke / Anomalous: Corns of smoke. Target words (" Bs ") were kept constant across conditions, while the first word in the phrase (" As ") was balanced for length and frequency. 135 students rated the stimuli for the major psycholinguistic features (concreteness, difficulty, familiarity, meaningfulness, cloze probability and aptness) in a paper-and-pencil test. 27 subjects (mean age 23 ± 3 years) took part in the ERP experiment. Brain activity (64 Ag/AgCl scalp electrodes, earlobes references, 250 Hz sampling rate) was recorded while subjects silently read the phrases (each word: 400 ms, interword interval: 200 ms) and performed a semantic-association task. After each phrase, a word pair was displayed, and subjects were instructed to choose the word that better matched the previous phrase. ERPs were time locked to the onset of the target word (" Bs "), while reaction times were measured …