Short-term semantic retention during on-line sentence comprehension. Brain potential evidence from filler-gap constructions

In filler-gap sentences, a phrase (‘filler’) is separated by intervening words from a subsequent phrase (‘gap’) with which it is integrated. The filler-gap interval provides a useful model for the study of short-term retention processes during sentence comprehension. Kluender and Kutas [J. Cogn. Neurosci. 5 (1993) 29] used event-related brain potentials (ERPs) to show that a filler phrase places a demand on short-term retention processes in the filler-gap interval, but left the processing level at which this demand arises unspecified. Here we use ERPs to address the issue of whether the filler places a demand on the semantic component of short-term retention processes in the filler-gap interval. Participants read filler-gap sentences, which began with a filler phrase and, in the filler-gap interval, contained a subject and object that were either semantically related or unrelated. There was also a control condition in which the filler phrase was absent (i.e. less memory demand). The main result was that during the filler-gap interval, bilateral posterior electrodes displayed a larger positivity for unrelated than related words. Moreover, during the same interval, posterior electrodes displayed a larger positivity for filler-gap sentences than for control sentences. In the control condition (non-filler gap sentences), manipulation of semantic relatedness did not produce differences in ERP activity. Our results suggest that a filler phrase places a demand on the semantic component of verbal working memory during on-line sentence comprehension.