Neurolinguistic research utilizing event-related brain potentials (ERPs) typically relates syntactic phrase structure processing to an early automatic processing stage around 150 to 200 msec, whereas morphosyntactic processing is associated with a later and somewhat more attention-dependent processing stage between 300 and 500 msec. However, recent studies have challenged this position by reporting highly automatic ERP effects for morphosyntax in the 100 to 200 msec time range. The present study aimed at determining the factors that could contribute to such shifts in latency and automaticity. In two experiments varying the degree of attention, German phrase structure and morphosyntactic violations were compared in conditions in which the locality of the violated syntactic relation, as well as the violation point and the acoustic properties of the speech stimuli, were strictly controlled for. A negativity between 100 and 300 msec after the violation point occurred in response to both types of syntactic violations and independently of the allocation of attentional resources. These findings suggest that the timing and automaticity of ERP effects reflecting specific syntactic subprocesses are influenced to a larger degree by methodological than by linguistic factors, and thus, need to be regarded as relative rather than fixed to temporally successive processing stages.

The present study investigated the automaticity of morphosyntactic processes and processes of syntactic structure building using event-related brain potentials. Two experiments were conducted, which contrasted the impact of local subject-verb agreement violations (Experiment 1) and word category violations (Experiment 2) on the mismatch negativity, an early event-related brain potential component reflecting automatic auditory change detection. The two violation types were realized in two-word utterances comparable with regard to acoustic parameters and structural complexity. The grammaticality of the utterances modulated the mismatch negativity response in both experiments, suggesting that both types of syntactic violations were detected automatically within 200 msec after the violation point. However, the topographical distribution of the grammaticality effect varied as a function of violation type, which indicates that the brain mechanisms underlying the processing of subject-verb agreement and word category information may be functionally distinct even at this earliest stage of syntactic analysis. The findings are discussed against the background of studies investigating syntax processing beyond the level of two-word utterances.