An fMRI study of syntactic layers: sentential and lexical aspects of embedding

Sentences with embedding are more complex than sentences without embedding, because they contain more syntactic layers in their phrasal architecture. Until now, neuroimaging studies tested embedded sentences that also included syntactic movement. To explore which cortical areas are specifically involved in the processing of syntactic layers, we used embedded sentences that did not include syntactic movement. We compared sentences in which the verb was complemented with an embedded sentence (CP complement) to sentences with noun phrase (NP) complements. The comparison of sentences with and without embedding revealed activations in the left IFG, bilateral temporo-parietal cortices, bilateral MTG, and precuneus. We further examined the effect of embedding at the lexical level. For this, we tested whether verbs that can select CP complements, namely, whose lexical entry includes a CP option, are more complex than verbs that do not select CP complements, even when both verb types appeared in sentences without embedding. The comparison of verbs that can take CP complements with verbs that can take NP complements, when these verbs appeared with prepositional phrase complements, revealed activations in bilateral anterior MTG and precuneus, showing that embedding also affects processing at the lexical level. The results thus show that sentential embedding, even without syntactic movement, loads onto several cortical resources including the left IFG, indicating that the generation of syntactic layers is cognitively costly, and that embedding affects processing at both the sentence and the lexical levels.