Brain areas underlying retrieval of nouns and verbs: grammatical class and task demand effects

Introduction Current data on the neural correlates of noun and verb processing are inconsistent as studies using different imaging techniques and/or different tasks have provided remarkably different results. This fMRIstudy was designed to clarify this issue by assessing task-independent and task-dependent grammatical-class-specific effects during lexical retrieval tasks. Materials and Methods - Subjects: 12 right-handed Italian students (F=6, M=6), aged 20-34, 15 years of schooling. Materials: subjects retrieved nouns and verbs during two different tasks: a picture naming task similar to the one adopted in previous experiments (PNT) and a retrieval task (NVRT ) were verbs or nouns were derived from the corresponding noun or verb (say “vincere” (to win) when reading “vittoria” (victory), or vicversa; NVRT, see Crepaldi et al., 2006). It should be noted that the typical picture naming task used to assess noun or verb retrieval is necessarily imbalanced for imageability and visual complexity, with pictures for verbs being more complex than pictures for objects. On the other hand, in NVRT task nouns and verbs had the same level of imageability. We also had a simple reading task for nouns or for verbs as a control experiment. Before fMRI scans, subject familiarized with the experimental stimuli so that strategic search of targets was minimized. Here we recorded VOTs for the different tasks. - fMRI data acquisition and analysis: 1.5 T Marconi-Philips Infinion Scanner, EPI gradient echo sequence TR: 3”; TE: 60 msec; Flip angle: 90°. Data were collected in alternating 30” blocks of baseline and experimental tasks. Condition-specific effects were estimated according to the general linear model (GLM) in SPM2. Statistical thresholds: p<0.001. Results Behavioural results. RT for nouns and verbs differed in a task dependent manner, so that VOTs were longer for verbs in the PNT and for nouns in the NVRT (task by grammatical class interaction significant at p<0.05). fMRI results. In both experimental tasks, verbs were consistently associated with an increased activation of the postcentral gyrus bilaterally, of the right superior parietal lobule and of the precuneus. No consistent network emerged for nouns across tasks. There was also a task dependent difference between verbs and nouns consistent with the behavioural data, with a greater activation of the left inferior frontal gyrus for verbs in the PNT, and for nouns in the NVRT. Discussion The systematic involvement of parietal cortices during verb processing suggests an intimate relationship between lexical-semantic and sensori-motor representations for actions. On the other hand, our data show that involvement of Broca’s area in lexical retrieval may not reflect a grammatical class effect (i.e., verbs vs nouns), rather it may be explained merely by task difficulty or task specific demands (see Thompson-Schill, 1997).