Electrophysiological analysis of spoken sentence processing as a function of working memory requirements

The processing of 216 naturally spoken sentences, especially of sentences containing relative clauses of differing complexities, was investigated by an event-related potential (ERP) analysis with 24 participants. The ERP data provided information both on the timing and the intensity of processes involved in spoken language comprehension. The onset and offset of different relative clause types is visible in the ERP waveforms, and the complexity of the relative clause modulates the amplitude of a slow negative potential over frontal and central electrode sites. The time course of this latter effect reflects that of the expected demands made on working memory, and its scalp distribution is consistent with neural generators in the frontal and temporal cortex which are suggested to be vitally important for working memory function during language processing. This interpretation is also supported by sentence comprehension data in that the subjects who showed better comprehension also showed larger sentence type effects. We thus have evidence that differences in sentence complexity can have electrophysiologically measurable effects on working memory usage during spoken language processing.