Neural mechanisms underlying noun-verb distinction- fMri evidence from semantic processing of chinese words

Abstract

Background. Numerous studies have been carried out to investigate whether and how grammatical class distinction may be represented in the human brain in the past three decades. However, as concluded by two recent extensive reviews (Crepaldi et al., 2011; Vigliocco et al., 2010), compelling evidence supporting separate neural correlates of noun and verb processing is still lacking, due to confounding from the frequent involvement of inflectional operations in experimental tasks and the fundamental differences between nouns and verbs that were not controlled for. In the current study, these limitations were circumvented by studying the comprehension and production of both concrete and abstract nouns and verbs at the semantic level in a language with minimal inflection morphology, i.e., Chinese. Methods. A semantic relatedness judgment task and a semantic associate production task were administered to the same group of 20 Mandarin-speaking participants, who were right handed, and aged from 19 to 29. For each experiment, imaging data were modeled by the four event types (concrete noun, concrete verb, abstract noun, abstract verb) at the subject level. Contrast maps of overall nouns versus verbs and vice versa were built, and fed into a 2 (noun, verb) *2 (judgment, production) within-subject design at the group level. Within the mask of positive activation for each word class (voxelwise p > 0.05), a conjunction analysis of grammatical contrasts between the two tasks was conducted with a corrected clusterwise significance of 0.05 (voxelwise p < 0.005, k = 49) to look for word class-specific regions that were independent of tasks. An ROI analysis of interaction effects between concreteness and word class was performed in each of the activated regions for the two tasks respectively, to examine possible differential grammatical class effects across concreteness levels. Results. The conjunction analysis revealed verb-specific activation in left posterior superior and middle temporal cortex (LpSTG&MTG, [-48, -51, 9], k = 124) and left inferior frontal area (LIFG, [-51, 6, 9], k = 49), and noun-specific activation in left fusiform gyrus ([-30, -30, -24], k = 96).No interaction effects were found in the ROI analyses for either experiment in the verbspecific regions (see figure 1). However, a significant interaction in the judgment task and a marginally significant interaction in the production task were observed in left fusiform gyrus, due to a reduced difference between abstract nouns and verbs. Conclusions. Through employing a language with impoverished inflectional morphology, and presenting abstract and concrete nouns and verbs in bare forms in two tasks involving semantic processing, we have identified the LpSTG&MTG and LIFG with greater activation for verbs, and left fusiform for nouns. However, the observation that the difference in neural response between nouns and verbs was greater for concrete than abstract items in left fusiform raises the possibility that noun-specific activation originated from higher imageability of concrete nouns than verbs. Nevertheless, verb preference in LpSTG&MTG and LIFG was not modulated by concreteness, unambiguously demonstrating distinctive neural correlates of verb processing even in languages with minimal inflection.