Tracing the interplay between syntactic and lexical features: fMRI evidence from agreement comprehension.

The current fMRI study was designed to investigate whether the processing of different gender-related cues embedded in nouns affects the computation of agreement dependencies and, if so, where this possible interaction is mapped in the brain. We used the Spanish gender agreement system, which makes it possible to manipulate two different factors: the agreement between different sentence constituents (i.e., by contrasting congruent versus incongruent determiner-noun pairs) and the formal (i.e., orthographical/morphological) and/or lexical information embedded in the noun -i.e., by contrasting transparent (e.g., libromasc. [book]; lunafem. [moon]) and opaque nouns (e.g., lápizmasc. [pencil]; vejezfem. [old age]). Crucially, these data illustrated, for the first time, how the network underlying agreement is sensitive to different gender-to-ending cues: different sources of gender information associated with nouns affect the neural circuits involved in the computation of local agreement dependencies. When the gender marking is informative (as in the case of transparent nouns), both formal and lexical information is used to establish grammatical relations. In contrast, when no formal cues are available (as in the case of opaque nouns), gender information is retrieved from the lexicon. We demonstrated the involvement of the posterior MTG/STG, pars triangularis within the IFG, and parietal regions during gender agreement computation. Critically, in order to integrate the different available information sources, the dynamics of this fronto-temporal loop change and additional regions, such as the hippocampus, the angular and the supramarginal gyri are recruited. These results underpin previous neuroanatomical models proposed in the context of both gender processing and sentence comprehension. But, more importantly, they provide valuable information regarding how and where the brain's language system dynamically integrates all the available form-based and lexical cues during comprehension.