OS_38. Language comprehension
Sunday, October 02nd, 2011 [15:40 - 16:40]
TALKS
OS_38.1 - Neural basis of semantic and syntactic interference resolution in sentence comprehension
Guo, Y. 1 , Martin, R. 1 , Hamilton, A. C. 1 , Van Dyke, J. A. 2 & Tan, Y. 1
1 Department of Psychology, Rice University, Houston, TX, USA
2 Haskins Laboratories, New Haven, CT, USA
Interference in sentence comprehension has been observed when a noun phrase that intervenes between a long distance subject-verb dependency has semantic or syntactic features that make it an appropriate subject of the verb (e.g., “customer” causes both semantic and syntactic interference in “The hostess who the customer ignored was talking loudly” because “customer” is a subject noun that is semantically plausible as the agent of talking). In a 2 (low vs. high semantic interference) × 2 (low vs. high syntactic interference) fMRI study, activation was measured during sentence and comprehension question processing to determine the brain regions involved in resolving semantic and syntactic interference (p corrected by cluster threshold < 0.01). One region in the left inferior frontal gyrus (LIFG; BA45) showed greater activation for the high than the low syntactic interference conditions following the onset of sentences. Interestingly, a largely overlapping region in the LIFG (BA45) also showed greater activation for the high than the low semantic interference conditions, but following the onset of comprehension questions. The results suggest that a region in the LIFG is involved in the resolution of both semantic and syntactic interference during sentence comprehension, but at different time points in processing.
OS_38.2 - The rostroventral-caudodorsal gradient of the caudate nucleus at 7T
Mestres Missé, A. , Turner, R. & Friederici, A. D.
Max Planck Institute for Human Cognitive and Brain Sciences
Lateral prefrontal cortex and basal ganglia work together to mediate working memory and top-down regulation of cognition. This system regulates the balance and interactions between automatic and high-order control responses. Using ultra-high-field high-resolution functional magnetic resonance imaging (fMRI), the present study examined the role of subcortical structures in cognitive control during language processing in humans. Participants were asked to judge the grammaticality of ambiguous, unambiguous and ungrammatical sentences. Unambiguous sentences require an automatic response, while ambiguous and ungrammatical sentences conflict with the automatic response and, hence, require a high-order control response. Furthermore, ambiguity and ungrammaticality represent two different dimensions of conflict resolution, while for ambiguity a correct alternative is available, that is not the case for ungrammaticality. The results reveal a rostroventral-caudodorsal axis in the head of the caudate nucleus with more rostroventral regions supporting higher levels of cognitive processing. This functional architecture mirrors the rostrocaudal hierarchical organization within the prefrontal cortex.
OS_38.3 - Interference and working memory in sentence comprehension
Tan, Y. & Martin, R.
Psychology Department, Rice University, Houston, USA
During sentence processing, interference effects have been obtained when comprehenders have to retrieve earlier information to link with later information across intervening material with features that partially match the target information. For example, for the sentence, “The resident who the manager evicted complained loudly,” interference is obtained from “manager” when retrieving the subject of “complained” because “manager” is a syntactic subject and is semantically plausible as the agent of “complained.” The working memory resources involved in retrieval and interference resolution were investigated using an individual differences approach. Ninety-six subjects read sentences in a self-paced, phrase-by-phrase fashion followed by a comprehension question. The degree of semantic and syntactic interference in the intervening material was manipulated factorially. During sentence processing, semantic interference effects were negatively correlated with semantic retention capacity, even after partialling out vocabulary knowledge. Syntactic interference during sentence processing was related only to reading span, a complex measure which involves syntactic processing. A measure of phonological retention capacity was found to be unrelated to any of the interference effects. The results are consistent with a multiple capacities approach to verbal working memory that includes separable phonological, semantic, and syntactic components, with only the latter two being relevant for sentence comprehension.