Poster E27, Saturday, August 18, 3:00 – 4:45 pm, Room 2000AB

Producing sentences in the MRI scanner: Effects of lexicality and verb arguments

Atsuko Takashima1, Antje Meyer1, Peter Hagoort1,2, Kirsten Weber1,2;1Max Planck Institute for Psycholinguistics, Nijmegen, the Netherlands, 2Donders Institute for Brain, Cognition and Behaviour, Nijmegen, the Netherlands

Neuroimaging studies have shown that a core network of inferior frontal cortex and posterior temporal cortex with a left hemisphere dominance is involved in sentence comprehension as well as production. Most sentence-level production studies in the MR scanner have focused on simple transitive sentence and we know little about the production of sentences featuring verbs differing in argument structure. Here we tested whether brain activation patterns, as measured with BOLD-fMRI, differed when participants were producing sentences which varied in the number of verb arguments (intransitive/transitive/ditransitive verbs). Furthermore, we assessed the effect of lexicality using existing Dutch verbs and pseudoverbs. This was done because processing parts of sentence structure, such as verb-arguments could be lexically-driven. We obtained functional MR images with a 3T scanner while 30 right-handed native speakers of Dutch produced sentences cued by visual input. The stimuli consisted of 18 written existing verbs and 18 pseudoverbs. These (pseudo)verbs were categorised to be used in intransitive, transitive or ditransitive sentences (6 verbs each - verb argument effect). Three geometric figures (circle, triangle, and square) were used as the nouns filling the roles of the agent and verb arguments. Trials were presented in mini-blocks per sentence structure condition. Each block started with an example for the type of structure to be produced (e.g. "The triangle gives the square to the circle"). On each trial, participants first saw the (pseudo)verb to-be-used in the centre of the screen for 1.5 sec, followed by the three geometric shapes in a horizontal layout for 5 sec. They were instructed to overtly produce a sentence with the same structure as the example, using the (pseudo)verb and the figures. Overall, a verb lexicality effect (verb > pseudoverb sentences) was found in the left inferior frontal gyrus (LIFG) and the left posterior middle temporal gyrus (MTG). Additionally, a more posterior bilateral middle temporal region extending to the angular gyrus showed greater activation during the verb than pseudoverb sentences. A verb argument effect (ditransitive > intransitive) was accompanied by an increase in activation in the left posterior MTG and bilateral precuneus. The left posterior MTG cluster overlapped substantially with the cluster found for the lexicality contrast. Thus, as in single word comprehension, producing sentences showed activation of the core language network in LIFG and left posterior MTG. This was especially the case when the verb was an existing word and the sentence had multiple argument slots to be filled. In summary, we show that the production network overlaps with the comprehension network. As for the lexicality effect, existing words compared to pseudowords have a memory representation with rich lexical-syntactic information activating the core language network during production. Retrieval of more complex lexical-syntactic patterns for sentences with more verb arguments lead to increased activation in the posterior language network related to the lexical-syntactic information stored in the mental lexicon.

Topic Area: Grammar: Syntax