The Role of N 400 Component of Event-Related Potentials in Subconscious Syntactic Processing and Conscious Semantic Representation (original) (raw)

A. SPECIFIC AIMS The syntax of a human language is a set of abstract rules which constitute a computational system that defines the grammatical sentences of a language regardless of their semantic content; these sentences are intuitively well-formed to a native speaker (Chomsky, 1965; Batterink and Neville, 2013). Syntax is the study of linguistic form and grammatical structure, and semantics is the study of meaning and reference in language use (Chomsky, 1975). Electroencephalographic (EEG) studies suggest that syntactical processing may use implicit learning mechanisms and occur below the level of conscious awareness (Batterink and Neville, 2013). The biphasic event-related potential (ERP) associated with syntactical violations consists of an early negativity and a late positivity (Batterink and Neville, 2013). A cross-modal distraction task, during which an auditory tone is paired with a visually presented syntactical violation, revealed that only consciously detected syntactic violations elicit both phases of the ERP; undetected syntactical violations exhibited no late positivity (Batterink and Neville, 2013). On the semantic side, a recent study by Frankland and Greene (2015) claimed to identify subregions of the left mid-superior temporal cortex (lmSTC) that encode distinct abstract semantic variables (e.g., "Who did it?" and "To whom was it done?") using functional magnetic resonance imaging (fMRI). We are interested in characterizing the relationship between subconscious syntactical processing and semantic processing. Theoretical linguists following Chomsky (1957, 1965, 1975) have argued for a distinction between syntax and semantics on the basis of sentences such as "Colorless green ideas sleep furiously," which is grammatical but meaningless. Frankland and Greene (2015), however, suggest that meaning is structure-dependent on the basis of such sentences as "The baby was kicked by the grandfather" and "The grandfather was kicked by the baby;" they also cite downstream afferent processing differences associated with differences in semantic variable position. We propose that metaphor is an apt domain for cognitive neuroscience to parse the relation between syntactical and semantical processing. As a comparison of two semantically distinct domains, nominal metaphors involve the figurative use of a noun, as in "Time is a thief" (Cardillo et al., 2010). Nominal metaphors can be transformed between the active and passive voice, a procedure which conserves its syntactic deep structure (Chomsky, 1965). However, it is unclear whether metaphoric meaning is conserved in the active-passive transformation (e.g. "Time is a thief" vs. "A thief is time"). If sentence meaning is structure dependent, the question of where meaning is derived from (deep or surface structure) is of much interest to theoretical linguists and cognitive neuroscientists. Work in the Chatterjee lab, based on findings by Bowdle and Gentner (2005), has shown that the process by which certain words take on additional and directly-accessible figurative meanings involves a qualitative shift in cognitive processing from comparison to categorization: Novel metaphor comprehension initially involves right-hemisphere semantic processing, but as metaphors become familiar, comprehension seems to be increasingly mediated by the left hemisphere (Cardillo et al., 2012). Based on work by Batterink and Neville (2013), Batterink et al. (2010), Frankland and Greene (2015), and the Chatterjee lab, we hypothesize that the N400 response to unconsciously detected syntactic violations (Batterink and Neville, 2013) and consciously detected semantic violations (Batterink et al., 2010) results from a failure to map semantic knowledge onto unconsciously generated syntactic structures. Aim 1.1 Determine baseline ERP components of active-and passive-voice sentence structure. The goal of this study is to see if meaningful differences in ERP waveform exist between processing active-and passive-voice sentence structures. Protocol from Experiment 1 of Frankland and Greene, 2015 will be replicated, substituting EEG of the left hemisphere for fMRI. We will use a 64-channel ActiveTwo system for ERP recordings (Batterink and Neville, 2013). This study will set a baseline ERP waveform for comparison in additional studies. Aim 1.2 Determine ERP component changes of symmetric and asymmetric metaphor constituents vs. literal active-and passive-voice sentences. This study will explore how the position of semantic variables in grammatical and ungrammatical metaphors affects ERP waveform, and whether any changes in waveform index the locations of semantic or syntactic constituents. To parse the relation between syntactical structure and semantic constituent position, nominal symmetric metaphors ("The cat is a thief" ⇒ "The thief is a cat"), nominal asymmetric metaphors ("The lawyer is a shark" ⇒ "The shark is a lawyer"), and active-and passive-voice transformed literal sentences will be presented to subjects as EEG is recorded. Aim 2. Track effects of conventionalization of metaphor on semantic and syntactic expectancy. This study will attempt to index changes in ERP waveform to structure-mapping failures. Using protocol developed by Cardillo et al. (2012), participants will be taught nominal metaphors to three levels of conventionalization. Subjects will be presented with the metaphors during EEG recording, with either syntactic subject-verb agreement violations or semantic violations.