Do Noun and Verb Processing Really Recruit Spatially-Segregated Neural Circuits?

The functional independence of noun and verb lexical retrieval has been repeatedly demonstrated over the last decade (e.g., Rapp and Caramazza, 2002); however, the evidence produced so far on whether these processes recruit separate neural circuits is far less clear (e.g., Bedny & Thompson‑Schill, 2006; Berlingeri et al., 2008). This may be due to the lack of homogeneity in the tasks used in the various experiments. The meta‑analysis described in this study aims at disentangling the brain regions that are systematically associated with a given grammatical class from those whose grammatical‑class specificity is modulated by the task used. We collected 620 activation coordinates associated with either nouns or verbs in simple effect analyses from 22 neuroimaging studies published from 1996 to 2008. A hierarchical cluster algorithm was employed adopting the Ward (1963) criterion to automatically segregate groups of coordinates into separate clusters (mean standard deviation < 7.5 mm in the x, y, and z directions for each cluster; Jobard et al., 2003). This procedure produced a set of 49 clusters, which were assessed with a binomial test for specificity for grammatical class and/or task, and with a Fisher test for task‑by‑grammatical class interaction. Two clusters turned out to be associated with nouns across different tasks, while five were associated with verbs (Figure 1a); the locations of these clusters do not support spatially segregated neural circuits for nouns and verbs, as suggested in some previous work (e.g., Cappa & Perani, 2003). Five clusters showed instead a task‑by‑grammatical class interaction (Figure 1b), which was mainly driven by noun‑specific activation in derivational tasks and by verb‑specific activation in picture naming and fluency tasks. These results will be discussed in the light of the cognitive processes entailed by the individual experimental tasks.