In the current study event-related brain potentials (ERPs) were
recorded to auditorily presented sentences in order to examine the
time course and topographic distribution of word-category
information and morpho-syntactic information during sentence
processing. Recent research reported that the violation of
word-category information elicited a particularly early anterior
negativity often lateralized to the left hemisphere (ELAN, 150-200
ms) while the violation of morpho-syntactic information elicited a
negativity with a similar distribution that occurred somewhat later
relative to the former violation (LAN, 300-500 ms). This was taken
to reflect two successive syntactic processing stages: a first
stage during which word-category information is processed and a
second - later - stage during which morpho-syntactic (and semantic)
information is processed. However, there has never been a direct
comparison of the two processes. The present study, therefore,
presented German native speakers with sentences that carried either
a word-category violation or a subject-verb-agreement violation.
The auditory materials were designed such that the different types
of violating information were provided at the same point in time
relative to the onsets of the critical words, namely not before the
final suffix of those words (e.g., (A) correct: `Die Sauce, die der
Koch verfeinert, schmeckt sehr lecker.', (B) agreement violation:
`Die Sauce, die der Koch[3rd Ps. Sg]. verfeinerst[2nd Ps. Sg]. ,
schmeckt sehr lecker', (C) word-category violation: `Die Sauce, die
der Koch als NP[verfeinertV], schmeckt sehr lecker'). The data
suggest that both types of violations elicit negativities that
started equally early in time (at about 270 ms after the critical
words' onsets). However, the distribution of those negative
components differed between the conditions: while word-category
violations (i.e., condition C) elicited a negativity that was most
pronounced at the outer left anterior electrodes, the agreement
violations (i.e., condition B) elicited an anterior negativity that
was more centrally distributed. With respect to timing, it
therefore appears that syntactic information is processed as soon
as it is accessible irrespective of the type of information. By
contrast, with respect to the topographic distribution of the
respective ERP-components, it appears that the two information
types can clearly be separated, suggesting different neural
generators for the processing of word category and morpho-syntactic
agreement information.