A role for left inferior frontal and posterior superior temporal cortex in extracting a syntactic tree from a sentence

Abstract

On reading the sentence “the kids who exhausted their parents slept”, how do we decide that it is the kids who slept and not the parents? The present behavioral and functional magnetic resonance imaging (fMRI) study explored the processes underlying the extraction of syntactically organized information from sentences. Participants were presented with sentences whose syntactic complexity was manipulated using either a center-embedded or an adjunct structure. The goal was to vary separately the sentence syntactic structure and the linear distance between the main verb and its subject. Each sentence was followed by a short subject + verb probe, and the participants had to check whether or not it matched a proposition expressed in the sentence. Behavioral and fMRI data showed a significant cost and enhanced activity within left inferior frontal and posterior superior temporal cortex whenever participants processed center-embedded sentences, which required extracting a nontrivial subtree formed by nonadjacent words. This syntactic complexity effect was not observed during online sentence processing but rather during the processing of the probe and only when the verification could not rely on a superficial lexical analysis. Moreover, the manipulation of linear distance affected performance and brain activity mainly when the sentences did not have a center-embedded structure. We did not find evidence suggesting that tree-extraction, a fundamental operation of a core syntax network, takes place during sentence comprehension. The present finding showed that the syntactic complexity effect, which is an outcome of this operation, became detectable later on, whenever we need to extract structural information not obvious in the superficial sequence of words.