EEG-oscillations during sentence processing

The project aims both at a linguistic and at an electrophysiological aspect. The linguistic aspect concerns the experimental investigation of two basically different processes involved in sentence processing. One process is a purely syntactic one. It controlls the serialization and it is determined by the grammar of the clause structure. The other process is a lexical-semantic one. The core experiments are the following two: Experiment 1 investigates the syntactically governed serialization during sentence production. The test subjects are faced with the following task: Sentences, presented word by word, must be syntactically rearranged. In experiment 2, the same stimulus material is used for a semantically determined substitution-by-retrieval task: An item in a clause must be substituted by its hyperonym. The leading idea is to get a grip on the differentiation of syntactically and semantically driven processing routines. In order to achieve this, various additional, control experiments that are described in the proposal will be necessary. According to findings of leason studies and imaging techniques, focal neural acticvities of the respective processes are concentrated in different cortical areas (frontal vs. temporal). So, each experiment should reveal different specific focal activities. The electrophysiological aspect concerns the fact that sentence-processing studies are predominatly designed as ERP-studies (event related potentials). Oscillation measurement techniques have hitherto been neglected, which is a disadvantage: Spectral analyses and related methods allow to capture different but simultaneously active cognitive processes in well-defined, distinct frequency bands. By a series of experiments (conducted by the team of the co-applicant) it was established that oscillations in the upper alpha band primarily correlate with the cognitive processing of a semantic stimulus. Oscillations in the lower alpha band, however, are primarily indicative of temporal shifts of attention. Thus, we expect for experiment 2 - as a semantic tasks (but not for the syntactic task in experiment 1) - a strong event-related desynchronization in the upper alpha band. In the absence of established results for syntactic processing, we hypothesize to find faster oscillations (e.g. in the range of the beta band), because syntactic processing is a less distributed phenomenon and is expected to take place in a more specific network component in comparison to semantic processing. The correlate in the power spectrum ought to be an increase in frequency. An additional aim is the comparison of the results of this investigation with the results of an fMRI-study (DFG- funded project, co-conducted by the applicant). Both studies use identical stimulus material. The combination of an EEG-method and an imaging technique is supposed to provide better insight both into topography as well as the spectral correlates of cognitive activities.

The project investigated the following question: Is it possible to identify domain specific cognitive processing components in the EEG-spektrum? The specific cognitive domain is language, in particular, sentence production. The declared aim of the project was the identification relevant EEG-correlates of basic language processing activities by means of spectral- and frequency-analytical methods. These methods investigate the EEG-records for correlates of specific processing activities in specific frequency bands. We hoped to detect different frequency- specific activities for syntactic processes and for lexical search and retrieval processes. The experiments were designed to isolate two basically different language specific processes: a purely syntactic process of structure generation and a lexical-semantic process (retrieval of a semantically identified expression). The applied method concentrates on the investigation of band power changes, measured during the execution of the experimentally triggered language processes. An essential goal of the experiments was to show that this novel approach can be applied fruitfully to the inquiry of language processing activities (as specialized cognitive processes). In particular, we expected different but frequency specific oscillations to show up according to the processing task which subsequently can be filtered from the total spectral activity and assigned to the respective task-specific portions. This basic aspect of the chosen paradigm could be strikingly validated in the experiments carried out. The experiments did not only reproduce the by now well known unspecific band power changes (general attention- and expectancy effects, working memory demands), but proved also successful with respect to the hypothesized specific band power changes. The main results are: Specifically linguistic cognitive processing activities do have correlates in the oscillation patterns. On the one hand, the upper alpha desynchronization (10-12 Hz.) correlates with semantic processing. On the other hand, semantic processing does not draw on general working memory capacities (no theta-band specific synchronizations). These results pave the way for further investigations in this area.